技术领域technical field
本发明涉及数据处理技术领域，尤其涉及一种面向用户的体征信息动态监护方法和动态监护系统。The invention relates to the technical field of data processing, in particular to a user-oriented dynamic monitoring method and a dynamic monitoring system for sign information.
背景技术Background technique
多参数床旁监护仪是一种常用的临床医疗设备。这种监护设备的特点是具有多组传感器，可以同时监测心电、血压、血氧、脉搏、呼吸、体温等生命体征指标。在病房监护中，多参数监护仪可以成为医生的对病人监护的重要参考，使医生能及时发现病人出现的问题并及时进行处理，从而保证了病人的生命安全。监护仪的临床应用可见于：手术中、手术后、外伤护理、冠心病、危重病人、新生儿、早产儿、高压氧舱、分娩室等。Multi-parameter bedside monitor is a commonly used clinical medical equipment. This monitoring device is characterized by multiple sets of sensors that can simultaneously monitor vital signs such as ECG, blood pressure, blood oxygen, pulse, respiration, and body temperature. In ward monitoring, the multi-parameter monitor can become an important reference for doctors to monitor patients, so that doctors can find out the problems of patients in time and deal with them in time, thus ensuring the safety of patients' lives. The clinical application of the monitor can be found in: operation, post operation, trauma care, coronary heart disease, critically ill patients, newborns, premature infants, hyperbaric oxygen chamber, delivery room, etc.
然而，在很多情况下，非住院人群也存在体征监测的需求，比如面向在家休养的患者、体弱需要进行实时体征监测人群、以及养老院/休养所等社会保障和福利机构的配套保障。目前业内并没有相应的设备以满足其使用需求。However, in many cases, non-hospitalized populations also have a need for sign monitoring, such as for patients recuperating at home, frail people who need real-time sign monitoring, and supporting guarantees for social security and welfare institutions such as nursing homes/retirement homes. At present, there is no corresponding equipment in the industry to meet its usage needs.
因此，如何面向非住院人群进行有效的体征监护，以及基于体征监护为用户提供更有效的医疗保障服务，是本发明所要解决的难题和挑战。Therefore, how to perform effective sign monitoring for non-hospitalized populations and provide users with more effective medical security services based on sign monitoring are the problems and challenges to be solved by the present invention.
发明内容Contents of the invention
本发明的目的是提供一种面向用户的体征信息动态监护方法和动态监护系统，通过有线或无线通讯技术实现数据交互，通过对心电图数据进行完整快速的自动分析和记录，能够及时发现异常并产生报警，当上述处理在服务器中完成时，还可以自动下发报警信息到动态监护设备，当上述处理在动态监护设备完成时，可以自动上报或由本地输出手动触发上报报警信息到服务器。此外，服务器能够基于报警信息进行信息分发处理，包括分发给医疗机构或者分发给被监测者的关联用户的终端设备。动态监护设备及服务器还都支持监护数据的记录和存储，并可配置为对全部数据记录以及仅对异数据记录的不同模式。The purpose of the present invention is to provide a user-oriented dynamic monitoring method and dynamic monitoring system for physical sign information, which can realize data interaction through wired or wireless communication technology, and can detect abnormalities in time and generate Alarm, when the above processing is completed in the server, it can also automatically send the alarm information to the dynamic monitoring device. When the above processing is completed in the dynamic monitoring device, it can be automatically reported or manually triggered by the local output to report the alarm information to the server. In addition, the server can perform information distribution processing based on the alarm information, including distribution to medical institutions or to terminal devices of associated users of the monitored person. Both the dynamic monitoring device and the server also support the recording and storage of monitoring data, and can be configured to record all data and record different modes only for different data.
为实现上述目的，本发明实施例第一方面提供了一种面向用户的体征信息动态监护方法，包括：In order to achieve the above purpose, the first aspect of the embodiment of the present invention provides a user-oriented method for dynamic monitoring of vital sign information, including:
动态监护设备接收用户输入或者服务器下发的监测基准数据；所述监测基准数据包括被测对象信息和心电异常事件信息；The dynamic monitoring device receives the monitoring reference data input by the user or issued by the server; the monitoring reference data includes the measured object information and the abnormal ECG event information;
所述动态监护设备对被测对象进行监护数据采集，得到所述被测对象的体征监护数据；所述体征监护数据至少包括心电图数据；The dynamic monitoring device collects the monitoring data of the measured object, and obtains the monitoring data of the physical signs of the measured object; the monitoring data of the physical signs at least includes electrocardiogram data;
对所述心电图数据进行波群特征识别，得到所述心电图数据的特征信号，根据所述特征信号对所述心电图数据进行心搏分类，结合心电图基本规律参考数据得到心搏分类信息，并生成心电图事件数据；所述心电事件数据包括所述动态监护设备的设备ID信息；Perform wave group feature recognition on the electrocardiogram data to obtain characteristic signals of the electrocardiogram data, perform heartbeat classification on the electrocardiogram data according to the characteristic signals, obtain heartbeat classification information in combination with the basic law reference data of the electrocardiogram, and generate an electrocardiogram Event data; the ECG event data includes the device ID information of the dynamic monitoring device;
根据所述心电图事件数据实时确定对应的心电图事件信息，并确定所述心电图事件信息是否为所述心电异常事件信息；Determine corresponding electrocardiogram event information in real time according to the electrocardiogram event data, and determine whether the electrocardiogram event information is the abnormal electrocardiogram event information;
当为所述心电异常事件信息时，生成并通过所述动态监护设备输出报警信息；所述报警信息包括所述设备ID信息、所述被测对象信息、所述心电异常事件信息；When it is the abnormal ECG event information, generate and output alarm information through the dynamic monitoring device; the alarm information includes the device ID information, the measured object information, and the abnormal ECG event information;
确定所述动态监护设备的工作模式；所述工作模式包括本地处理模式和后台处理模式；Determine the working mode of the dynamic monitoring device; the working mode includes a local processing mode and a background processing mode;
当为本地处理模式时，所述动态监护设备对所述心电图事件数据和所述报警信息进行记录存储；When in the local processing mode, the dynamic monitoring device records and stores the electrocardiogram event data and the alarm information;
当为后台处理模式时，所述服务器根据所述报警信息，或者根据接收到所述动态监护设备基于所述报警信息触发的报警信号，确定所述被测对象信息对应的责任用户的用户ID，并生成通知信息发送给所述责任用户的用户设备和/或预设的关联机构的用户设备；其中，所述通知信息携包括所述报警信息和所述责任用户的用户ID。When it is in the background processing mode, the server determines the user ID of the responsible user corresponding to the measured object information according to the alarm information, or according to the received alarm signal triggered by the dynamic monitoring device based on the alarm information, And generate notification information and send it to the user equipment of the responsible user and/or the user equipment of the preset associated organization; wherein, the notification information includes the alarm information and the user ID of the responsible user.
优选的，在所述动态监护设备接收用户输入或者服务器下发的监测基准数据之前，所述方法还包括：Preferably, before the dynamic monitoring device receives the monitoring benchmark data input by the user or sent by the server, the method further includes:
确定所述被测对象的被测对象信息。Determine the measured object information of the measured object.
优选的，所述方法还包括：Preferably, the method also includes:
所述服务器接收所述责任用户的用户设备发送的反馈信息，并根据所述设备ID信息将所述反馈信息发送给所述动态监护设备；The server receives the feedback information sent by the user equipment of the responsible user, and sends the feedback information to the dynamic monitoring equipment according to the equipment ID information;
所述动态监护设备输出所述反馈信息。The ambulatory monitoring device outputs the feedback information.
优选的，所述生成通知信息发送给所述责任用户的用户设备和/或预设的关联机构的用户设备具体包括：Preferably, the sending of the generating notification information to the user equipment of the responsible user and/or the user equipment of the preset associated organization specifically includes:
所述服务器根据所述被测对象信息和/或设备ID信息，确定所述动态监护设备的位置信息；The server determines the location information of the dynamic monitoring device according to the measured object information and/or device ID information;
根据所述位置信息和所述报警信息，生成所述通知信息；generating the notification information according to the location information and the alarm information;
根据所述责任用户的用户ID将所述通知信息发送给所述责任用户的用户设备和/或预设的关联机构的用户设备。Sending the notification information to the user equipment of the responsible user and/or the user equipment of a preset associated organization according to the user ID of the responsible user.
优选的，所述方法还包括：Preferably, the method also includes:
所述动态监护设备按照预设时间信息，将预设时段内的心电图数据发送给所述服务器；The dynamic monitoring device sends the electrocardiogram data within a preset time period to the server according to preset time information;
所述服务器根据所述心电图数据生成预设时段内所述被测对象的监测报告数据。The server generates monitoring report data of the measured object within a preset time period according to the electrocardiogram data.
优选的，所述方法还包括：Preferably, the method also includes:
所述动态监护设备接收用户触发的事件记录信号，监听并生成报警事件记录信息；The dynamic monitoring device receives the event record signal triggered by the user, monitors and generates alarm event record information;
将所述报警事件记录信息发送给所述责任用户的用户设备和/或预设的关联机构的用户设备。Sending the alarm event record information to the user equipment of the responsible user and/or the user equipment of a preset associated institution.
优选的，所述体征监护数据还包括：脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据中的一种或多种。Preferably, the sign monitoring data further includes: one or more of pulse data, blood pressure data, respiration data, blood oxygen saturation data and body temperature data.
进一步优选的，所述方法还包括：Further preferably, the method also includes:
所述动态监护设备确定所述脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据中的一个或多个存在超出相应的设定阈值的异常数据，并结合所述心电图事件数据生成体征监测异常事件信息，并发送到所述服务器。The dynamic monitoring device determines that one or more of the pulse data, blood pressure data, respiration data, blood oxygen saturation data and body temperature data has abnormal data exceeding the corresponding set threshold, and combines the electrocardiogram event data to generate The signs monitor abnormal event information and send it to the server.
优选的，所述对所述心电图数据进行波群特征识别，得到所述心电图数据的特征信号，根据所述特征信号对所述心电图数据进行心搏分类，结合心电图基本规律参考数据得到心搏分类信息，并生成心电图事件数据具体包括：Preferably, performing wave group feature recognition on the electrocardiogram data to obtain the characteristic signal of the electrocardiogram data, performing heartbeat classification on the electrocardiogram data according to the characteristic signal, and obtaining the heartbeat classification in combination with the basic law reference data of the electrocardiogram information, and generate ECG event data specifically including:
将所述心电图数据的数据格式经过重采样转换为预设标准数据格式，并对转换后的预设标准数据格式的心电图数据进行第一滤波处理；Converting the data format of the electrocardiogram data into a preset standard data format through resampling, and performing a first filtering process on the converted electrocardiogram data in a preset standard data format;
对所述第一滤波处理后的心电图数据进行心搏检测处理，识别所述心电图数据包括的多个心搏数据，每个所述心搏数据对应一个心搏周期，包括相应的P波、QRS波群、T波的幅值和起止时间数据；Perform heartbeat detection processing on the electrocardiogram data after the first filtering process, identify a plurality of heartbeat data included in the electrocardiogram data, each of the heartbeat data corresponds to a heartbeat cycle, including corresponding P waves, QRS Amplitude and start and end time data of wave group and T wave;
根据所述心搏数据确定每个心搏的检测置信度；determining a detection confidence level for each heartbeat based on the heartbeat data;
根据干扰识别二分类模型对所述心搏数据进行干扰识别，得到心搏数据是否存在干扰噪音，以及用于判断干扰噪音的一个概率值；Perform interference identification on the heart beat data according to the interference identification binary classification model to obtain whether there is interference noise in the heart beat data and a probability value for judging the interference noise;
根据所述检测置信度确定心搏数据的有效性，并且，根据确定有效的心搏数据的导联参数和心搏数据，基于所述干扰识别的结果和时间规则合并生成心搏时间序列数据；根据所述心搏时间序列数据生成心搏分析数据；Determine the validity of the heartbeat data according to the detection confidence, and, according to the lead parameters and the heartbeat data that determine the valid heartbeat data, combine the results of the interference identification and time rules to generate heartbeat time series data; generating heartbeat analysis data according to the heartbeat time series data;
根据心搏分类模型对所述心搏分析数据进行幅值和时间表征数据的特征提取和分析，得到所述心搏分析数据的一次分类信息；performing feature extraction and analysis on the amplitude and time representation data of the heartbeat analysis data according to the heartbeat classification model, to obtain primary classification information of the heartbeat analysis data;
对所述一次分类信息结果中的特定心搏的心搏分析数据输入到ST段和T波改变模型进行识别，确定ST段和T波评价信息；Input the heartbeat analysis data of a specific heartbeat in the first classification information result into the ST segment and T wave change model for identification, and determine the ST segment and T wave evaluation information;
根据所述心搏时间序列数据，对所述心搏分析数据进行P波和T波特征检测，确定每个心搏中P波和T波的详细特征信息，详细特征信息包括幅值、方向、形态和起止时间的数据；According to the heartbeat time series data, the P wave and T wave feature detection is performed on the heartbeat analysis data, and the detailed feature information of the P wave and T wave in each heartbeat is determined. The detailed feature information includes amplitude, direction, data on morphology and start and end times;
对所述心搏分析数据在所述一次分类信息下根据所述心电图基本规律参考数据、所述P波和T波的详细特征信息以及所述ST段和T波评价信息进行二次分类处理，得到心搏分类信息；Perform secondary classification processing on the heartbeat analysis data under the primary classification information according to the basic rule reference data of the electrocardiogram, the detailed feature information of the P wave and T wave, and the ST segment and T wave evaluation information, Obtain heartbeat classification information;
对所述心搏分类信息进行分析匹配，生成所述心电图事件数据。Analyzing and matching the cardiac beat classification information to generate the electrocardiogram event data.
本发明实施例提供的面向用户的体征信息动态监护方法，采用数据的预处理，心搏特征检测，基于深度学习方法的干扰信号检测和心搏分类与导联合并，心搏的审核，心电图事件和参数的分析计算，最终自动输出心电事件结果数据的一个完整快速流程的自动化心电检测分析，并且可基于心电检测分析结果输出报警，或结合血压、血氧、脉搏、呼吸、体温数据产生报警，以及基于报警的响应处理，通过基于报警信息进行信息分发处理，包括分发给医疗机构或者分发给被监测者的关联用户的终端设备，使得被监测者得到有效、及时的医疗救助服务。本发明的面向用户的体征信息动态监护方法，面向非住院人群进行有效的体征监护，并基于体征监护为用户提供更有效的医疗保障服务。The user-oriented dynamic monitoring method for physical sign information provided by the embodiment of the present invention adopts data preprocessing, heartbeat feature detection, interference signal detection based on deep learning methods, heartbeat classification and lead combination, heartbeat review, and electrocardiogram events. and parameter analysis and calculation, and finally automatically output the ECG event result data of a complete and fast process of automatic ECG detection and analysis, and can output alarms based on the ECG detection and analysis results, or combine blood pressure, blood oxygen, pulse, respiration, body temperature data Alarms are generated, and alarm-based response processing is performed. Information distribution processing based on alarm information, including distribution to medical institutions or terminal devices of associated users of the monitored person, enables the monitored person to receive effective and timely medical assistance services. The user-oriented dynamic monitoring method for sign information of the present invention performs effective sign monitoring for non-hospitalized populations, and provides users with more effective medical security services based on the sign monitoring.
本发明实施例第二方面提供了一种动态监护系统，该系统包括上述第一方面所述的一个或多个动态监护设备和服务器。The second aspect of the embodiments of the present invention provides an ambulatory monitoring system, which includes one or more ambulatory monitoring devices and a server described in the first aspect.
附图说明Description of drawings
图1为本发明实施例提供的面向用户的体征信息动态监护方法的流程图；Fig. 1 is a flowchart of a user-oriented method for dynamic monitoring of vital sign information provided by an embodiment of the present invention;
图2为本发明实施例提供的心电图数据的处理方法的流程图；Fig. 2 is the flow chart of the processing method of electrocardiogram data provided by the embodiment of the present invention;
图3为本发明实施例提供的干扰识别二分类模型的示意图；FIG. 3 is a schematic diagram of a binary classification model for interference identification provided by an embodiment of the present invention;
图4为本发明实施例提供的心搏分类模型的示意图；FIG. 4 is a schematic diagram of a heart beat classification model provided by an embodiment of the present invention;
图5为本发明实施例提供的ST段和T波改变模型的示意图；5 is a schematic diagram of the ST segment and T wave change model provided by the embodiment of the present invention;
图6为本发明实施例提供的一种动态监护系统的结构示意图。Fig. 6 is a schematic structural diagram of a dynamic monitoring system provided by an embodiment of the present invention.
具体实施方式Detailed ways
下面通过附图和实施例，对本发明的技术方案做进一步的详细描述。The technical solutions of the present invention will be described in further detail below with reference to the accompanying drawings and embodiments.
本发明提出的动态监护系统，面向非入院患者进行多参数体征监护。不同于临床监护，本发明的动态监护系统所涉及的数据传输的实时性、数据传输、存储和异常的上报及响应机制都与临床监护有着本质区别。The dynamic monitoring system proposed by the present invention performs multi-parameter sign monitoring for non-hospitalized patients. Different from clinical monitoring, the real-time data transmission, data transmission, storage and abnormal reporting and response mechanisms involved in the dynamic monitoring system of the present invention are essentially different from clinical monitoring.
此外，通过对现有监护仪的研究我们发现，在检测的心电、血压、血氧、脉搏、呼吸、体温等生命体征指标中，心电的监测是不同于其他各项参数的，通过传感器得到的心电信号需要通过一系列复杂的算法计算才能提取出其中的有效信息，相对其他信号而言处理过程比较复杂困难，也是容易出现检测错误的环节。In addition, through the research on the existing monitors, we found that among the vital sign indicators detected such as ECG, blood pressure, blood oxygen, pulse, respiration, body temperature, etc., the monitoring of ECG is different from other parameters. The obtained ECG signal needs to be calculated through a series of complex algorithms to extract the effective information. Compared with other signals, the processing process is more complicated and difficult, and it is also a link that is prone to detection errors.
心电信号是心肌细胞的电活动在体表反映出的微弱电流，通过体表电极和放大记录系统记录下来。在记录过程中同时还会记录到其他非心源性的电信号，比如骨骼肌活动带来的肌电信号干扰等等。因此我们认为需要对心电信号进行有效的干扰识别和排除，才能够有效降低因为干扰信号造成的误报。ECG signal is the weak current reflected by the electrical activity of cardiomyocytes on the body surface, which is recorded through body surface electrodes and an amplifying recording system. During the recording process, other non-cardiac electrical signals are also recorded, such as EMG interference caused by skeletal muscle activity. Therefore, we believe that it is necessary to effectively identify and eliminate the interference of the ECG signal in order to effectively reduce the false positives caused by the interference signal.
此外，心电信号是心肌电活动过程的体现，因此心电信号除了可以用来检测心率以外，还可以体现出大量的心脏状态的信息。在心脏状态出现问题的时候，心电信号会出现相应的改变。在对业内现有的多参数监护设备进行研究的过程中我们发现，现有的监护设备只能对心电信号进行非常有限的分析和报警。对此，除了对心电信号进行有效的干扰识别和排除，以降低因为干扰信号造成的误报之外，我们认为还可以从以下几点进行改进：In addition, the electrocardiographic signal is the embodiment of the electrical activity process of the myocardium, so the electrocardiographic signal can not only be used to detect the heart rate, but also reflect a large amount of information about the state of the heart. When there is a problem with the state of the heart, the ECG signal will change accordingly. In the process of researching the existing multi-parameter monitoring equipment in the industry, we found that the existing monitoring equipment can only perform very limited analysis and alarm on the ECG signal. In this regard, in addition to effective interference identification and elimination of ECG signals to reduce false alarms caused by interference signals, we believe that improvements can be made from the following points:
第一，在心搏特征提取中需要对P波、T波进行准确识别，可以避免心搏检测的多检和漏检，比如对一些特殊心电图信号，例如心律比较缓慢患者的高大T波，或者T波肥大的信号的多检。First, it is necessary to accurately identify P waves and T waves in heartbeat feature extraction, which can avoid multiple detection and missed detection of heartbeat detection, such as some special ECG signals, such as tall T waves in patients with slow heart rhythms, or T waves Multiple detection of wave hypertrophy signals.
第二，对心搏的分类进行更加细致的划分，而不能仅停留在窦性、室上性和室性这三种分类，从而满足临床心电图医生复杂全面的分析要求。Second, the classification of heartbeats should be classified more carefully, instead of just staying in the three classifications of sinus, supraventricular and ventricular, so as to meet the complex and comprehensive analysis requirements of clinical electrocardiography doctors.
第三，准确识别房扑房颤和ST-T改变，从而能够有助于提供对ST段和T波改变对心肌缺血分析的帮助。Third, accurate identification of atrial flutter and atrial fibrillation and ST-T changes can help to provide help for the analysis of ST segment and T wave changes on myocardial ischemia.
第四，对心搏和心电事件的准确识别。Fourth, accurate identification of heartbeat and ECG events.
在本发明中，我们针对上述几点，通过对心电数据的分析计算，特别是引入人工智能(AI)技术，对采集的数字信号进行心律失常分析、长间歇停搏，扑动和颤动，传导阻滞，早搏和逸搏，心动过缓，心动过快，ST段改变检测、心电事件的分析与归类，以达到产生准确报警信号的目的，从而有效的进行病人生命体征的监护。In the present invention, we aim at the above points, by analyzing and calculating the electrocardiographic data, especially introducing artificial intelligence (AI) technology, arrhythmia analysis, long intermittent arrest, flutter and tremor are carried out to the digital signal collected, Conduction block, premature beat and escape beat, bradycardia, tachycardia, ST segment change detection, analysis and classification of ECG events, in order to achieve the purpose of generating accurate alarm signals, so as to effectively monitor the patient's vital signs.
有数据表明，90％以上的心脏疾病突发都是在医疗机构之外发生的，因此对于有心脏疾病史的人群，记录和监控日常状态下的心脏情况，是非常有必要的。Statistics show that more than 90% of heart disease emergencies occur outside medical institutions. Therefore, it is very necessary for people with a history of heart disease to record and monitor their heart conditions in their daily state.
基于上述发现，本发明提出了一种面向用户的体征信息动态监护方法，可以应用于医疗机构外的面向用户的监护。可以实现于各种动态监护设备中，包括可穿戴式设备。下面结合图1所示的面向用户的体征信息动态监护方法的流程图，对本发明的面向用户的体征信息动态监护方法进行详述。在本发明的体征信息动态监护中，对心电信号的监护又是其中最重要的。Based on the above findings, the present invention proposes a user-oriented method for dynamic monitoring of vital sign information, which can be applied to user-oriented monitoring outside medical institutions. It can be implemented in various ambulatory monitoring devices, including wearable devices. The user-oriented dynamic monitoring method for vital sign information of the present invention will be described in detail below in conjunction with the flow chart of the user-oriented dynamic monitoring method for vital sign information shown in FIG. 1 . In the dynamic monitoring of sign information of the present invention, the monitoring of electrocardiographic signals is the most important.
如图1所示，本发明的面向用户的体征信息动态监护方法主要包括如下步骤：As shown in Figure 1, the user-oriented dynamic monitoring method for physical sign information of the present invention mainly includes the following steps:
步骤110，动态监护设备接收用户输入或者服务器下发的监测基准数据；Step 110, the dynamic monitoring device receives the monitoring benchmark data input by the user or sent by the server;
具体的，动态监护设备具体可以是包括有单导联或多导联心电监测功能的多参数监测设备，每台动态监护设备都有唯一的设备ID。当动态监护设备被分派给一个待监测用户使用时，可以根据该用户的情况，在动态监护设备中配置相应的监测基准数据。因此，需要首先确定被测对象的被测对象信息。Specifically, the dynamic monitoring device may specifically be a multi-parameter monitoring device including a single-lead or multi-lead ECG monitoring function, and each dynamic monitoring device has a unique device ID. When the dynamic monitoring device is assigned to a user to be monitored, corresponding monitoring benchmark data can be configured in the dynamic monitoring device according to the user's situation. Therefore, it is necessary to determine the measured object information of the measured object first.
对于心电监测来说，监测基准数据可以理解为用以指示所监测到的用户心电信号正常与否的是否需要产生报警的基准数据或信息，对于不同的用户，监测基准数据的设置可以不同，具体的可以通过在动态监护设备上配置输入的方式或者通过服务器根据用户信息进行配置并下发到动态监护设备的方式获得。在本实施例中，监测基准数据可以包括有被测对象信息和设定好的心电异常事件信息。心电异常事件信息包含有需要产生心电异常报警的各种心电异常事件的信息，在动态监护设备对心电图数据进行采集、分析等一系列处理，得到心电图数据指示的心电异常事件时，可以通过确定该心电异常事件是否是心电异常事件信息中规定的事件而确定是否产生报警。For ECG monitoring, the monitoring reference data can be understood as the reference data or information used to indicate whether the monitored user's ECG signal is normal or not, and whether an alarm needs to be generated. For different users, the settings of the monitoring reference data can be different , specifically, it can be obtained by configuring input on the dynamic monitoring device or by configuring the server according to user information and sending it to the dynamic monitoring device. In this embodiment, the monitoring reference data may include measured object information and set abnormal ECG event information. The abnormal ECG event information includes information on various abnormal ECG events that need to generate abnormal ECG alarms. When the dynamic monitoring equipment collects and analyzes the ECG data and obtains abnormal ECG events indicated by the ECG data, Whether to generate an alarm can be determined by determining whether the abnormal ECG event is an event specified in the abnormal ECG event information.
对于其他体征参数，比如脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据，监测基准数据可以是设的相应的参数阈值。每个参数的参数阈值都可以有不同的多组参数阈值，可以根据被监测者的实际情况进行选择。优选的，在本发明中，可以在监测前预先根据被测对象确定监测基准数据，然后根据监测基准数据确定相应的设定阈值。For other physical sign parameters, such as pulse data, blood pressure data, respiration data, blood oxygen saturation data, and body temperature data, the monitoring reference data can be set corresponding parameter thresholds. The parameter threshold of each parameter can have multiple groups of different parameter thresholds, which can be selected according to the actual situation of the monitored person. Preferably, in the present invention, the monitoring reference data can be determined in advance according to the measured object before monitoring, and then the corresponding setting threshold can be determined according to the monitoring reference data.
在本例中，监测基准数据至少包括被测对象信息和心电异常事件信息。In this example, the monitoring reference data includes at least the measured object information and the abnormal ECG event information.
步骤120，动态监护设备对被测对象进行监护数据采集，得到被测对象的体征监护数据；Step 120, the dynamic monitoring device collects the monitoring data of the measured object, and obtains the monitoring data of the measured object's signs;
具体的，动态监护设备具有与被测对象相接触的电极、探头、袖带等体征信号采集装置，通过体征信号采集装置采集被测对象的体征信号，并通过数字化处理得到体征监护数据。体征监护数据至少包括心电图数据，还可能包括上述所述的脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据等。体征监护数据具有时间属性信息，每个数据点都有对应的数据采集时间，这个时间即是时间属性信息。在进行数据采集的同时，这个数据采集时间也被纪录下来，并作为体征监护数据的时间属性信息进行存储。Specifically, the dynamic monitoring equipment has sign signal acquisition devices such as electrodes, probes, and cuffs that are in contact with the measured object. The sign signal acquisition device collects the sign signals of the measured object, and obtains sign monitoring data through digital processing. The sign monitoring data includes at least electrocardiogram data, and may also include the aforementioned pulse data, blood pressure data, respiration data, blood oxygen saturation data, and body temperature data. Sign monitoring data has time attribute information, and each data point has a corresponding data collection time, which is the time attribute information. While the data is being collected, the data collection time is also recorded and stored as the time attribute information of the sign monitoring data.
为更好地理解本发明的意图和实现方式，下面对各类体征监护数据的采集方法和原理进行简要介绍说明：In order to better understand the intent and implementation of the present invention, a brief introduction to the collection methods and principles of various sign monitoring data is given below:
心电图数据：通过无创心电图检查的心电信号采集记录仪队心脏细胞电生理活动产生的信号以单导联或多导联的形式进行采集记录。Electrocardiogram data: through non-invasive electrocardiogram examination, the electrocardiographic signal acquisition recorder collects and records the signals generated by the electrophysiological activities of cardiac cells in the form of single or multiple leads.
脉搏数据：脉搏是动脉血管随心脏舒缩而周期性博动的现象，脉搏包含血管内压、容积、位移和管壁张力等多种物理量的变化。我们优选的采用光电容积式脉搏测量，传感器由光源和光电变换器两部分组成，可夹在被测者的指尖或耳廓上。光源选择对动脉血中氧合血红蛋白有选择性的波长，比如采用光谱在700-900nm的发光二极管。这束光透过人体外周血管，当动脉充血容积变化时，改变了这束光的透光率，由光电变换器接收经组织透射或反射的光，转变为电信号送放大器放大和输出，由此反映动脉血管的容积变化。脉搏是随心脏的博动而周期性变化的信号，动脉血管容积也周期性地变化，光电变换器的信号变化周期就是脉搏率，即脉搏数据。Pulse data: Pulse is a phenomenon in which arterial blood vessels pulsate periodically with cardiac contraction, and pulse includes changes in various physical quantities such as intravascular pressure, volume, displacement, and vessel wall tension. We prefer to use photoelectric volumetric pulse measurement. The sensor is composed of a light source and a photoelectric transducer, which can be clamped on the fingertip or auricle of the person being measured. The light source selects a wavelength that is selective to oxyhemoglobin in arterial blood, for example, a light-emitting diode with a spectrum of 700-900nm is used. This beam of light passes through the peripheral blood vessels of the human body. When the volume of arterial congestion changes, the light transmittance of this beam of light is changed. The light transmitted or reflected by the tissue is received by the photoelectric converter and converted into an electrical signal, which is sent to the amplifier for amplification and output. This reflects the volume change of arterial blood vessels. The pulse is a signal that changes periodically with the beating of the heart, and the volume of the arteries also changes periodically. The signal change period of the photoelectric transducer is the pulse rate, that is, the pulse data.
血压数据：心脏收缩时所达到的最高压力称为收缩压，它把血液推进到主动脉，并维持全身循环。心脏扩张时所达到的最低压力称为舒张压，它使血液能回流到右心房。血压波形在一周内的积分除以心周期T称为平均压。血压数据的测量有多种方法可实现，具体可分为有创测量和无创测量。在多参数监护仪中我们优选采用柯氏音法和测振法两类无创测量方法。柯氏音法是检测袖带下的柯氏音(脉搏声)来测定血压的，柯氏音无创血压监护系统包括袖带充气系统、袖带、柯氏音传感器、音频放大及自动增益调整电路、A/D转换器、微处理器及显示部分等。测振法是检测气袖内气体的振荡波，振荡波源于血管壁的搏动，测量振荡波的相关点就可测定血压数据，包括收缩压(PS)，舒张压(PD)和平均压(PM)。测振法获得脉搏振动波的方法可借助微音器和压力传感器，通过测量得到脉搏振动波即得到血压数据。对于一些特殊的应用场景下，也可以通过有创测量的方式来获得血压数据。比如对于重症加强护理组(ICU)病房的一些病人进行监测，就可通过直接在动脉进行插管，将插管的另一端连接到消毒过的注满液体的压力检测系统中实现血压数据的实时采集。这种有创监测方法的优点包括：可以实时的显示出血压大小，并可以显示连续的血压变化波形；在低血压状态可以有准确的读数；长期记录的病人舒适度得到提升，避免无创测量中长期充气放气导致的创伤；可以提取出更多的信息，包括从血压波形的形态上可以推算出血管容量等。Blood pressure data: The highest pressure reached when the heart contracts is called the systolic pressure, which pushes blood into the aorta and maintains circulation throughout the body. The lowest pressure reached when the heart expands is called the diastolic pressure, which allows blood to flow back into the right atrium. The integral of the blood pressure waveform in one week divided by the cardiac cycle T is called the mean pressure. There are many ways to measure blood pressure data, which can be divided into invasive measurement and non-invasive measurement. In the multi-parameter monitor, we prefer to use two types of non-invasive measurement methods, the Korotkoff sound method and the vibration measurement method. The Korotkoff sound method is to detect the Korotkoff sound (pulse sound) under the cuff to measure blood pressure. The Korotkoff sound non-invasive blood pressure monitoring system includes a cuff inflation system, a cuff, a Korotkoff sound sensor, an audio amplifier and an automatic gain adjustment circuit. , A/D converter, microprocessor and display part, etc. The vibration measurement method is to detect the oscillating wave of the gas in the air cuff. The oscillating wave originates from the pulsation of the blood vessel wall. The blood pressure data can be measured by measuring the relevant points of the oscillating wave, including systolic blood pressure (PS), diastolic blood pressure (PD) and mean pressure (PM ). The method of obtaining the pulse vibration wave by means of vibration measurement can obtain the blood pressure data by measuring the pulse vibration wave with the help of a microphone and a pressure sensor. For some special application scenarios, blood pressure data can also be obtained through invasive measurement. For example, for the monitoring of some patients in the intensive care unit (ICU) ward, it is possible to directly intubate the artery and connect the other end of the cannula to a sterilized liquid-filled pressure detection system to realize real-time blood pressure data collection. The advantages of this invasive monitoring method include: it can display the blood pressure in real time, and can display the continuous blood pressure change waveform; it can have accurate readings in the state of hypotension; the comfort of the patient can be improved for long-term records, and it can avoid non-invasive measurement. Trauma caused by long-term inflation and deflation; more information can be extracted, including vascular volume can be deduced from the shape of the blood pressure waveform.
呼吸数据：呼吸测量是肺动能检查的重要部分。监护仪通过测量呼吸波来测定呼吸频率(次/分钟)，即得到呼吸数据。呼吸频率的测量可通过热敏电阻直接测量呼吸气流的温度变化，经过电桥电路将这一变化变换成电压信号；也可采用阻抗法来测量呼吸频率，因为呼吸运动时，胸壁肌肉交变张驰，胸廓交替变形，肌体组织的电阻抗也随之交替变化。测量呼吸阻抗值的变化可采用电桥法、调制法、恒压源法和恒流源法等多种方式。在监护仪中，呼吸阻抗电极亦可与心电电极合用，检测心电信号时可同时检测呼吸阻抗变化和呼吸频率。Respiration Data: Respiration measurements are an important part of the lung performance test. The monitor measures the respiratory frequency (times/minute) by measuring the respiratory wave, and obtains the respiratory data. The measurement of respiratory frequency can directly measure the temperature change of respiratory airflow through a thermistor, and convert this change into a voltage signal through a bridge circuit; the respiratory frequency can also be measured by impedance method, because the chest wall muscles alternately relax during breathing movement , the thorax deforms alternately, and the electrical impedance of the body tissue also changes alternately. Various methods such as bridge method, modulation method, constant voltage source method and constant current source method can be used to measure the change of respiratory impedance value. In the monitor, the respiratory impedance electrode can also be used together with the ECG electrode, and the respiratory impedance change and the respiratory frequency can be detected at the same time when detecting the ECG signal.
血氧饱和度数据：血氧饱和度是衡量人体血液携带氧的能力的重要参数。血氧饱和度的测量可以采用透射法(或反射法)双波长(红光R和红外光IR)光电检测技术，检测红光和红外光通过动脉血的光吸收引起的交变成分之比和非脉动组织(表皮、肌肉、静脉血等)引起光吸收的稳定分量(直流)值，通过计算可得到血氧饱和度值SpO2,即血氧饱和度数据。由于光电信号的脉动规律与心脏搏动的规律一致，所以根据检出信号的周期亦可同时确定脉搏数据。Blood oxygen saturation data: blood oxygen saturation is an important parameter to measure the ability of human blood to carry oxygen. The measurement of blood oxygen saturation can adopt the transmission method (or reflection method) dual-wavelength (red light R and infrared light IR) photoelectric detection technology to detect the ratio and sum of the alternating components caused by the light absorption of red light and infrared light through arterial blood. The stable component (direct current) value of light absorption caused by non-pulsating tissues (epidermal, muscle, venous blood, etc.) can be calculated to obtain the blood oxygen saturation value SpO2, that is, the blood oxygen saturation data. Since the pulsation law of the photoelectric signal is consistent with the heart beat law, the pulse data can also be determined at the same time according to the period of the detected signal.
体温数据：体温是了解生命状态的重要指标。体温的测量采用负温度系数的热敏电阻作为温度传感器，采用电桥作为检测电路。我们在具体的应用中可以采用集成化测温电路进行测量得到体温数据。亦可使用两道以上的测温电路，测量两个不同部位的温差对测量值进行修正。还可以采用体表探头和体腔探头，分别监护体表和腔内温度。在一些特殊的应用中，为了避免交叉传染，亦可以采用红外非接触测温技术来进行提问数据的监测。在监护仪中，我们设定测温精度在0.1℃，以便有较快的测温响应。Body temperature data: body temperature is an important indicator to understand the state of life. The measurement of body temperature uses a thermistor with a negative temperature coefficient as a temperature sensor, and a bridge as a detection circuit. In specific applications, we can use an integrated temperature measurement circuit to measure and obtain body temperature data. It is also possible to use more than two temperature measuring circuits to measure the temperature difference of two different parts to correct the measured value. Body surface probes and body cavity probes can also be used to monitor body surface and cavity temperatures, respectively. In some special applications, in order to avoid cross-infection, infrared non-contact temperature measurement technology can also be used to monitor the question data. In the monitor, we set the temperature measurement accuracy at 0.1°C in order to have a faster temperature measurement response.
在本发明中，我们可以通过上述方法对被测对象进行体征监护数据采集，得到被测对象的体征监护数据。In the present invention, we can collect the sign monitoring data of the measured object through the above method, and obtain the sign monitoring data of the measured object.
在前面已经提到，心电的监测相对于其他血压、血氧、脉搏、呼吸、体温等生命体征指标的监测更为复杂，因此在本发明中对于心电图数据采用了不同于其他体征监护数据的处理方法，具体采用基于人工智能自学习的心电图自动分析方法进行心电图数据的识别、处理和异常判断。As mentioned above, the monitoring of ECG is more complicated than the monitoring of other vital signs indicators such as blood pressure, blood oxygen, pulse, respiration, body temperature, etc. The processing method specifically adopts the electrocardiogram automatic analysis method based on artificial intelligence self-learning to identify, process and abnormally judge the electrocardiogram data.
在下述流程中也是主要针对心电图数据的处理过程为例进行体征信息动态监护方法的说明。其他体征数据可作为参考信息，与心电图数据的处理结果相结合，用以进行被监测者体征状态的判定。In the following process, the method of dynamic monitoring of sign information is also explained mainly by taking the processing process of electrocardiogram data as an example. Other sign data can be used as reference information, combined with the processing results of the electrocardiogram data, to determine the sign status of the monitored person.
步骤130，对心电图数据进行波群特征识别，得到心电图数据的特征信号，根据特征信号对心电图数据进行心搏分类，结合心电图基本规律参考数据得到心搏分类信息，并生成心电图事件数据；Step 130, performing wave group feature recognition on the electrocardiogram data to obtain the characteristic signal of the electrocardiogram data, performing heartbeat classification on the electrocardiogram data according to the characteristic signal, and obtaining heartbeat classification information in combination with the basic law reference data of the electrocardiogram, and generating electrocardiogram event data;
具体的，在本例中，对心电图数据的处理，可以在动态监护设备中执行，也可以是在服务器中执行。其中，动态监护设备可以通过有线方式或无线网络与服务器进行连接进行数据传输。优选的在进行实时数据传输时采用无线网络实现数据传输。这里所说的无线网络包括无线但不限于基于IEEE 802.11b标准的无线局域网(WIFI)，蓝牙，3G/4G/5G移动通信网络，物联网等方式。Specifically, in this example, the processing of the electrocardiogram data can be performed in the dynamic monitoring device, or can be performed in the server. Wherein, the dynamic monitoring device can be connected to the server through wired or wireless network for data transmission. Preferably, a wireless network is used to realize data transmission during real-time data transmission. The wireless network mentioned here includes but is not limited to wireless local area network (WIFI) based on IEEE 802.11b standard, Bluetooth, 3G/4G/5G mobile communication network, Internet of Things and other methods.
在动态监护设备监测得到心电图数据后，可以将心电图数据保存在动态监护设备本地，也可以是将数据传输至服务器中进行保存，在传输的数据中加载有动态监护设备的设备ID的信息。从而能够相应的对应得到被监测者的信息。After the dynamic monitoring device monitors and obtains the electrocardiogram data, the electrocardiogram data can be stored locally in the dynamic monitoring device, or the data can be transmitted to the server for storage, and the device ID information of the dynamic monitoring device is loaded in the transmitted data. In this way, the information of the monitored person can be correspondingly obtained.
本发明的心电图数据的处理过程，采用了基于人工智能自学习的心电图自动分析方法，是基于人工智能卷积神经网络(CNN)模型来实现的。CNN模型是深度学习中的监督学习方法，就是一个模拟神经网络的多层次网络(隐藏层hidden layer)连接结构，输入信号依次通过每个隐藏层，在其中进行一系列复杂的数学处理(Convolution卷积、Pooling池化、Regularization正则化、防止过拟合、Dropout暂时丢弃、Activation激活、一般使用Relu激活函数)，逐层自动地抽象出待识别物体的一些特征，然后把这些特征作为输入再传递到高一级的隐藏层进行计算，直到最后几层的全连接层(Full Connection)重构整个信号，使用Softmax函数进行逻辑(logistics)回归，达到多目标的分类。The processing procedure of electrocardiogram data of the present invention has adopted the electrocardiogram automatic analysis method based on artificial intelligence self-learning, is to realize based on artificial intelligence convolutional neural network (CNN) model. The CNN model is a supervised learning method in deep learning. It is a multi-level network (hidden layer) connection structure that simulates a neural network. The input signal passes through each hidden layer in turn, and a series of complex mathematical processing (Convolution Volume) is performed in it. Product, Pooling pooling, Regularization regularization, prevent overfitting, Dropout temporary discard, Activation activation, generally use Relu activation function), automatically abstract some features of the object to be recognized layer by layer, and then pass these features as input Go to the higher hidden layer for calculation, until the last few layers of the full connection layer (Full Connection) reconstruct the entire signal, and use the Softmax function to perform logistic regression to achieve multi-objective classification.
CNN属于人工智能中的监督学习方法，在训练阶段，输入信号经过多个的隐藏层处理到达最后的全连接层，softmax逻辑回归得到的分类结果，与已知的分类结果(label标签)之间会有一个误差，深度学习的一个核心思想就是通过大量的样本迭代来不断地极小化这个误差，从而计算得到连接各隐藏层神经元的参数。这个过程一般需要构造一个特别的损失函数(cost function)，利用非线性优化的梯度下降算法和误差反向传播算法(backpropagation algorithm，BP)，快速有效地极小化整个深度(隐藏层的层数)和广度(特征的维数)都十分复杂的神经网络结构中所有连接参数。CNN belongs to the supervised learning method in artificial intelligence. In the training phase, the input signal is processed through multiple hidden layers to reach the final fully connected layer. The classification result obtained by softmax logistic regression is compared with the known classification result (label label). There will be an error. A core idea of deep learning is to continuously minimize this error through a large number of sample iterations, so as to calculate the parameters connecting the neurons of each hidden layer. This process generally requires the construction of a special loss function (cost function), using the gradient descent algorithm of nonlinear optimization and the error backpropagation algorithm (backpropagation algorithm, BP), to quickly and effectively minimize the entire depth (the number of layers of the hidden layer ) and breadth (the dimensionality of features) are all connection parameters in the neural network structure that are very complex.
深度学习把需要识别的数据输入到训练模型，经过第一隐藏层、第二隐藏层、第三隐藏层，最后是输出识别结果。Deep learning inputs the data to be recognized into the training model, passes through the first hidden layer, the second hidden layer, and the third hidden layer, and finally outputs the recognition result.
在本发明中，对心电图数据进行波群特征识别、干扰识别、心搏分类等都是基于人工智能自学习的训练模型来得到输出结果，分析速度快，准确程度高。In the present invention, wave group feature identification, interference identification, heartbeat classification, etc. are all based on artificial intelligence self-learning training models to obtain output results for electrocardiogram data, and the analysis speed is fast and the accuracy is high.
无论是在动态监护设备中或者在服务器中执行对心电图数据的处理过程，其具体处理过程均可参考图3所示的流程，按照下述步骤进行，Whether it is in the dynamic monitoring device or in the server to perform the processing of the electrocardiogram data, its specific processing can refer to the process shown in Figure 3, and proceed according to the following steps,
步骤131，将心电图数据的数据格式经过重采样转换为预设标准数据格式，并对转换后的预设标准数据格式的心电图数据进行第一滤波处理；Step 131, converting the data format of the electrocardiogram data into a preset standard data format after resampling, and performing a first filtering process on the converted electrocardiogram data in the preset standard data format;
具体的，心电图数据的格式适配读取，对不同的设备有不同的读取实现，读取后，需要调整基线、根据增益转换成毫伏数据。经过数据重采样，把数据转换成全流程能够处理的采样频率。然后通过滤波去除高频，低频的噪音干扰和基线漂移，提高人工智能分析准确率。将处理后的心电图数据以预设标准数据格式保存。Specifically, the format of the electrocardiogram data is adapted for reading, and different devices have different reading implementations. After reading, the baseline needs to be adjusted and converted into millivolt data according to the gain. After data resampling, the data is converted into a sampling frequency that can be processed by the whole process. Then remove high-frequency and low-frequency noise interference and baseline drift through filtering to improve the accuracy of artificial intelligence analysis. Save the processed ECG data in a preset standard data format.
通过本步骤解决不同在使用不同导联，采样频率和传输数据格式的差异，以及通过数字信号滤波去除高频，低频的噪音干扰和基线漂移。Through this step, the differences in the use of different leads, sampling frequency and transmission data format are resolved, and the high frequency and low frequency noise interference and baseline drift are removed through digital signal filtering.
数字信号滤波可以分别采用高通滤波器，低通滤波器和中值滤波，把工频干扰、肌电干扰和基线漂移干扰消除，避免对后续分析的影响。Digital signal filtering can use high-pass filter, low-pass filter and median filter respectively to eliminate power frequency interference, myoelectric interference and baseline drift interference and avoid the impact on subsequent analysis.
更具体的，可以采用低通、高通巴特沃斯滤波器进行零相移滤波，以去除基线漂移和高频干扰，保留有效的心电信号；中值滤波则可以利用预设时长的滑动窗口内数据点电压幅值的中位数替代窗口中心序列的幅值。可以去除低频的基线漂移。More specifically, low-pass and high-pass Butterworth filters can be used for zero-phase-shift filtering to remove baseline drift and high-frequency interference and retain valid ECG signals; median filtering can use The median of the voltage magnitudes of the data points replaces the magnitude of the series in the center of the window. Low frequency baseline drift can be removed.
步骤132，对第一滤波处理后的心电图数据进行心搏检测处理，识别心电图数据包括的多个心搏数据；Step 132, performing heartbeat detection processing on the electrocardiogram data after the first filtering process, and identifying a plurality of heartbeat data included in the electrocardiogram data;
具体的，每个心搏数据对应一个心搏周期，包括相应的P波、QRS波群、T波的幅值和起止时间数据。本步骤中的心搏检测由两个过程构成，一是信号处理过程，从所述第一滤波处理后的心电图数据中提取QRS波群的特征频段；二是通过设置合理的阈值确定QRS波群的发生时间。在心电图中，一般会包含P波、QRS波群、T波成分以及噪声成分。一般QRS波群的频率范围在5到20Hz之间，可以通过一个在此范围内的带通滤波器提出QRS波群信号。然而P波、T波的频段以及噪声的频段和QRS波群频段有部分重叠，因此通过信号处理的方法并不能完全去除非QRS波群的信号。因此需要通过设置合理的阈值来从信号包络中提取QRS波群位置。具体的检测过程是一种基于峰值检测的过程。针对信号中每一个峰值顺序进行阈值判断，超过阈值时进入QRS波群判断流程，进行更多特征的检测，比如RR间期、形态等。Specifically, each heartbeat data corresponds to a heartbeat cycle, including corresponding P wave, QRS wave group, T wave amplitude and start and end time data. Heartbeat detection in this step is made up of two processes, the one, signal processing process, extract the characteristic frequency band of QRS wave group from the electrocardiogram data after described first filter processing; The 2nd, determine QRS wave group by setting reasonable threshold value time of occurrence. In an electrocardiogram, it generally includes P waves, QRS complexes, T wave components, and noise components. Generally, the frequency range of the QRS complex is between 5 and 20 Hz, and the QRS complex signal can be raised through a band-pass filter within this range. However, the frequency bands of P waves, T waves, and noises partially overlap with the frequency bands of QRS complexes, so signal processing methods cannot completely remove non-QRS complex signals. Therefore, it is necessary to extract the QRS complex position from the signal envelope by setting a reasonable threshold. The specific detection process is a process based on peak detection. Threshold judgment is performed for each peak sequence in the signal, and when the threshold is exceeded, it enters the QRS complex judgment process to detect more features, such as RR interval, shape, etc.
多参数监护仪往往是进行长时间记录，其过程中心搏信号的幅度和频率时时刻刻都在变化，并且在疾病状态下，这种特性会表现的更强。在进行阈值设定时，需要根据数据特征在时域的变化情况动态的进行阈值调整。为了提高检测的准确率和阳性率，QRS波群检测大多采用双幅度阈值结合时间阈值的方式进行，高阈值具有更高的阳性率，低阈值具有更高的敏感率，在RR间期超过一定时间阈值，使用低阈值进行检测，减少漏检情况。而低阈值由于阈值较低，容易受到T波、肌电噪声的影响，容易造成多检，因此优先使用高阈值进行检测。Multi-parameter monitors often record for a long time, and the amplitude and frequency of the heartbeat signal are changing all the time during the process, and this characteristic will be stronger in the disease state. When setting the threshold, it is necessary to dynamically adjust the threshold according to the change of the data characteristics in the time domain. In order to improve the accuracy and positive rate of detection, QRS complex detection is mostly carried out by double-amplitude threshold combined with time threshold. High threshold has a higher positive rate, and low threshold has a higher sensitivity rate. Time threshold, use a low threshold for detection to reduce missed detections. However, low thresholds are easily affected by T waves and EMG noise due to their low thresholds, and are likely to cause multiple detections. Therefore, high thresholds are preferred for detection.
对于不同导联的心搏数据都具有导联参数，用以表征该心搏数据为哪个导联的心搏数据。因此在得到心电图数据的同时也就可以根据其传输来源确定了其导联的信息，将此信息作为心博数据的导联参数。The heartbeat data of different leads have lead parameters, which are used to characterize the heartbeat data of which lead the heartbeat data is. Therefore, when the electrocardiogram data is obtained, the lead information can be determined according to its transmission source, and this information can be used as the lead parameter of the heartbeat data.
步骤133，根据心搏数据确定每个心搏的检测置信度；Step 133, determining the detection confidence of each heartbeat according to the heartbeat data;
具体的，置信度计算模块在心搏检测的过程中，根据QRS波群的幅度以及RR间期内噪声信号的幅度比例可以提供针对QRS波群检测置信度的估计值。Specifically, during the heartbeat detection process, the confidence calculation module can provide an estimated value for the confidence of the QRS complex detection according to the amplitude of the QRS complex and the amplitude ratio of the noise signal in the RR interval.
步骤134，根据干扰识别二分类模型对心搏数据进行干扰识别，得到心搏数据是否存在干扰噪音，以及用于判断干扰噪音的一个概率值；Step 134, perform interference identification on the heartbeat data according to the interference identification binary classification model, and obtain whether there is interference noise in the heartbeat data, and a probability value for judging the interference noise;
因为多参数监护仪在长时间记录过程中易受多种影响出现干扰现象，导致获取的心搏数据无效或不准确，不能正确反映受测者的状况，同时也增加医生诊断难度及工作量；而且干扰数据也是导致智能分析工具无法有效工作的主要因素。因此，将外界信号干扰降到最低显得尤为重要。Because the multi-parameter monitor is susceptible to various influences and interference phenomena during the long-term recording process, resulting in invalid or inaccurate heartbeat data, which cannot correctly reflect the condition of the subject, and also increases the difficulty and workload of doctors' diagnosis; Moreover, interference data is also the main factor that prevents intelligent analysis tools from working effectively. Therefore, it is particularly important to minimize external signal interference.
本步骤基于以深度学习算法为核心的端到端二分类识别模型，具有精度高，泛化性能强的特点，可有效地解决电极片脱落、运动干扰和静电干扰等主要干扰来源产生的扰动问题，克服了传统算法因干扰数据变化多样无规律而导致的识别效果差的问题。This step is based on the end-to-end binary classification recognition model with the deep learning algorithm as the core. It has the characteristics of high precision and strong generalization performance, and can effectively solve the disturbance problems caused by the main sources of interference such as electrode sheet falling off, motion interference and static interference. , which overcomes the problem of poor recognition effect caused by the traditional algorithm due to the variety and irregular changes of interference data.
具体可以通过如下方法来实现：Specifically, it can be achieved by the following methods:
步骤A，对心搏数据使用干扰识别二分类模型进行干扰识别；Step A, using the interference identification binary classification model to perform interference identification on the heartbeat data;
步骤B，识别心搏数据中，心搏间期大于等于预设间期判定阈值的数据片段；Step B, identifying data segments in the heartbeat data whose heartbeat interval is greater than or equal to a preset interval determination threshold;
步骤C，对心搏间期大于等于预设间期判定阈值的数据片段进行信号异常判断，确定是否为异常信号；Step C, performing signal abnormality judgment on the data segment whose heartbeat interval is greater than or equal to the preset interval judgment threshold, and determining whether it is an abnormal signal;
其中，异常信号的识别主要包括是否为电极片脱落、低电压等情况。Among them, the identification of abnormal signals mainly includes whether the electrodes are off, low voltage and so on.
步骤D，如果不是异常信号，则以预设时间宽度，根据设定时值确定数据片段中滑动取样的起始数据点和终止数据点，并由起始数据点开始对数据片段进行滑动取样，至终止数据点为止，得到多个取样数据段；Step D, if it is not an abnormal signal, determine the start data point and end data point of sliding sampling in the data segment according to the preset time width with the preset time width, and start sliding sampling of the data segment from the starting data point, Until the termination data point, a plurality of sampling data segments are obtained;
步骤E，对每个取样数据段进行干扰识别。Step E, performing interference identification on each sampled data segment.
以一个具体的例子对上述步骤A-E进行说明。对每个导联的心搏数据以设定的第一数据量进行切割采样，然后分别输入到干扰识别二分类模型进行分类，获得干扰识别结果和对应结果的一个概率值；对心搏间期大于等于2秒的心搏数据，先判断是否是信号溢出，低电压，电极脱落；如果不是上述情况，就按照第一数据量，从左边心搏开始，向右连续以第一数据量不重叠滑动取样，进行识别。A specific example is used to illustrate the above steps A-E. The heartbeat data of each lead is cut and sampled with the set first data volume, and then input into the interference identification binary classification model for classification, and a probability value of the interference identification result and the corresponding result is obtained; For heartbeat data greater than or equal to 2 seconds, first judge whether it is signal overflow, low voltage, or electrode drop-off; if it is not the above situation, start from the left heartbeat according to the first data amount, and continue to the right without overlapping the first data amount Sliding sampling for identification.
输入可以是任一导联的第一数据量心搏数据，然后采用干扰识别二分类模型进行分类，直接输出是否为干扰的分类结果，获得结果快，精确度高，稳定性好，可为后续分析提供更有效优质的数据。The input can be the first data amount of heartbeat data of any lead, and then use the interference recognition binary classification model to classify, and directly output the classification result of whether it is interference. The result is obtained quickly, with high accuracy and good stability, which can be used for subsequent Analysis provides more effective and high-quality data.
因为干扰数据往往是由外界扰动因素的作用而引起的，主要有电极片脱落、低电压、静电干扰和运动干扰等情况，不但不同扰动源产生的干扰数据不同，而且相同扰动源产生的干扰数据也是多种多样；同时考虑到干扰数据虽然多样性布较广，但与正常数据的差异很大，所以在收集干扰的训练数据时也是尽可能的保证多样性，同时采取移动窗口滑动采样，尽可能增加干扰数据的多样性，以使模型对干扰数据更加鲁棒，即使未来的干扰数据不同于以往任何的干扰，但相比于正常数据，其与干扰的相似度也会大于正常数据，从而使模型识别干扰数据的能力增强。Because the interference data is often caused by the action of external disturbance factors, mainly including electrode sheet falling off, low voltage, static interference and motion interference, etc., not only the interference data generated by different disturbance sources are different, but also the interference data generated by the same disturbance source It is also diverse; at the same time, considering that although the interference data has a wide diversity distribution, it is very different from the normal data, so when collecting the interference training data, it is also necessary to ensure diversity as much as possible. At the same time, sliding sampling with a moving window is adopted. It is possible to increase the diversity of interference data to make the model more robust to interference data. Even if future interference data is different from any previous interference, compared with normal data, its similarity to interference will be greater than normal data, thus Enhance the ability of the model to identify noisy data.
本步骤中采用的干扰识别二分类模型可以如图3所示，网络首先使用2层卷积层，卷积核大小是1x5,每层后加上一个最大值池化。卷积核数目从128开始，每经过一次最大池化层，卷积核数目翻倍。卷积层之后是两个全连接层和一个softmax分类器。由于该模型的分类数为2，所以softmax有两个输出单元，依次对应相应类别，采用交叉熵做为损失函数。The interference recognition binary classification model used in this step can be shown in Figure 3. The network first uses two convolutional layers, the convolution kernel size is 1x5, and a maximum pooling is added after each layer. The number of convolution kernels starts from 128, and the number of convolution kernels doubles every time a maximum pooling layer is passed. The convolutional layers are followed by two fully connected layers and a softmax classifier. Since the classification number of the model is 2, softmax has two output units, which correspond to the corresponding categories in turn, and cross-entropy is used as the loss function.
对于该模型的训练，我们采用了来源于30万病人近400万精确标注的数据片段。标注分为两类：正常心电图信号或者是有明显干扰的心电图信号片段。我们通过定制开发的工具进行片段标注，然后以自定义标准数据格式保存干扰片段信息。For the training of the model, we used nearly 4 million precisely labeled data segments from 300,000 patients. The annotations are divided into two categories: normal ECG signals or ECG signal fragments with obvious interference. We use custom-developed tools for fragment annotation, and then save interference fragment information in a custom standard data format.
在训练过程，使用两台GPU服务器进行几十次轮循训练。在一个具体的例子中，采样率是200Hz，数据长度是300个心电图电压值(毫伏)的一个片段D[300]，输入数据是：InputData(i，j),其中，i是第i个导联，j是导联i第j个片段D。输入数据全部经过随机打散才开始训练，保证了训练过程收敛，同时，控制从同一个病人的心电图数据中收集太多的样本，提高模型的泛化能力，既真实场景下的准确率。训练收敛后，使用100万独立的测试数据进行测试，准确率可以到达99.3％。另有具体测试数据如下表1。During the training process, dozens of round-robin trainings are performed using two GPU servers. In a specific example, the sampling rate is 200Hz, the data length is a segment D[300] of 300 electrocardiogram voltage values (millivolts), and the input data is: InputData(i, j), where i is the i-th Lead, j is the jth segment D of lead i. All input data are randomly scattered before training, which ensures the convergence of the training process. At the same time, it controls the collection of too many samples from the ECG data of the same patient, improving the generalization ability of the model, which is the accuracy rate in real scenarios. After the training converges, use 1 million independent test data for testing, and the accuracy rate can reach 99.3%. Other specific test data are shown in Table 1 below.
表1Table 1
步骤135，根据检测置信度确定心搏数据的有效性，并且，根据确定有效的心搏数据的导联参数和心搏数据，基于干扰识别的结果和时间规则合并生成心搏时间序列数据，并根据心搏时间序列数据生成心搏分析数据；Step 135, determine the validity of the heartbeat data according to the detection confidence, and, according to the lead parameters and heartbeat data that determine the valid heartbeat data, combine the results of interference identification and time rules to generate heartbeat time series data, and Generate heartbeat analysis data based on heartbeat time series data;
具体的，由于心电图信号的复杂性以及每个导联可能受到不同程度的干扰影响，依靠单个导联检测心搏会存在多检和漏检的情况，不同导联检测到心搏结果的时间表征数据没有对齐，所以需要对所有导联的心搏数据根据干扰识别结果和时间规则进行合并，生成一个完整的心搏时间序列数据，统一所有导联心搏数据的时间表征数据。其中，时间表征数据用于表示每个数据点在心电图数据信号时间轴上的时间信息。根据这个统一的心搏时间序列数据，在后续的分析计算时，可以使用预先设置好的阀值，对各导联心搏数据进行切割，从而生成具体分析需要的各导联的心搏分析数据。Specifically, due to the complexity of the ECG signal and the fact that each lead may be affected by different degrees of interference, relying on a single lead to detect heartbeats will have multiple detections and missed detections. The time representation of heartbeat results detected by different leads The data is not aligned, so it is necessary to merge the heartbeat data of all leads according to the interference identification results and time rules to generate a complete heartbeat time series data and unify the time representation data of all lead heartbeat data. Wherein, the time representation data is used to represent the time information of each data point on the time axis of the electrocardiogram data signal. According to this unified heartbeat time series data, in the subsequent analysis and calculation, the preset threshold can be used to cut the heartbeat data of each lead, so as to generate the heartbeat analysis data of each lead required for specific analysis .
上述每个导联的心搏数据在合并前，需要根据步骤133中获得的检测置信度确定心搏数据的有效性。Before merging the above-mentioned heartbeat data of each lead, it is necessary to determine the validity of the heartbeat data according to the detection confidence obtained in step 133 .
具体的，导联心搏合并模块执行的心搏数据合并过程如下：根据心电图基本规律参考数据的不应期获取不同导联心搏数据的时间表征数据组合，丢弃其中偏差较大的心搏数据，对上述时间表征数据组合投票产生合并心搏位置，将合并心搏位置加入合并心搏时间序列，移动到下一组待处理的心搏数据，循环执行直至完成所有心搏数据的合并。Specifically, the heartbeat data merging process performed by the lead heartbeat merging module is as follows: According to the refractory period of the reference data of the basic rules of the electrocardiogram, the time representation data combination of different lead heartbeat data is obtained, and the heartbeat data with large deviations are discarded , vote for the combination of the above time representation data to generate a merged heartbeat position, add the merged heartbeat position to the merged heartbeat time series, move to the next set of heartbeat data to be processed, and execute in a loop until the merger of all heartbeat data is completed.
其中，心电图活动不应期可以优选在200毫秒至280毫秒之间。获取的不同导联心搏数据的时间表征数据组合应满足以下条件：心搏数据的时间表征数据组合中每个导联最多包含一个心搏数据的时间表征数据。在对心搏数据的时间表征数据组合进行投票时，使用检出心搏数据的导联数占有效导联数的百分比来决定；若心搏数据的时间表征数据对应导联的位置为低电压段、干扰段以及电极脱落时认为该导联对此心搏数据为无效导联。在计算合并心搏具体位置时，可以采用心搏数据的时间表征数据平均值得到。在合并过程中，本方法设置了一个不应期来避免错误合并。Wherein, the refractory period of the electrocardiogram may preferably be between 200 milliseconds and 280 milliseconds. The acquired time representation data combinations of heartbeat data from different leads should meet the following conditions: each lead in the time representation data combination of heartbeat data contains at most one time representation data of heartbeat data. When voting on the combination of time representation data of heartbeat data, it is determined by the percentage of the number of leads with detected heartbeat data in the number of valid leads; if the position of the lead corresponding to the time representation data of heartbeat data is low voltage When segment, interference segment and electrode fall off, it is considered that the lead is an invalid lead for this heartbeat data. When calculating the specific location of the combined heartbeat, it can be obtained by using the average value of the time representation data of the heartbeat data. During merging, this method sets a refractory period to avoid erroneous merging.
在本步骤中，通过合并操作输出一个统一的心搏时间序列数据。该步骤同时能够降低心搏的多检率和漏检率，有效的提高心搏检测的敏感度和阳性预测率。In this step, a unified heartbeat time series data is output through the merge operation. At the same time, this step can reduce the over-detection rate and missed-detection rate of heartbeat, and effectively improve the sensitivity and positive prediction rate of heartbeat detection.
步骤136，根据心搏分类模型对心搏分析数据进行幅值和时间表征数据的特征提取和分析，得到心搏分析数据的一次分类信息；Step 136, perform feature extraction and analysis on the amplitude and time representation data of the heartbeat analysis data according to the heartbeat classification model, and obtain primary classification information of the heartbeat analysis data;
不同心电监测设备在信号测量、采集或者输出的导联数据等方面存在的差异，因此可以根据具体情况，使用简单的单导联分类方法，或者是多导联分类方法。多导联分类方法又包括导联投票决策分类方法和导联同步关联分类方法两种。导联投票决策分类方法是基于各导联的心搏分析数据进行导联独立分类，再把结果投票融合确定分类结果的投票决策方法；导联同步关联分类方法则采用对各导联的心搏分析数据进行同步关联分析的方法。单导联分类方法就是对单导联设备的心搏分析数据，直接使用对应导联模型进行分类，没有投票决策过程。下面对以上所述几种分类方法分别进行说明。Different ECG monitoring devices have differences in signal measurement, acquisition, or output lead data. Therefore, a simple single-lead classification method or a multi-lead classification method can be used according to specific conditions. The multi-lead classification method includes the lead voting decision classification method and the lead synchronous association classification method. The lead voting decision classification method is based on the heartbeat analysis data of each lead for independent classification of the leads, and then votes the results to determine the voting decision method of the classification results; the lead synchronous correlation classification method uses the heartbeat of each lead Methods for analyzing data for simultaneous association analysis. The single-lead classification method is to directly use the corresponding lead model to classify the heartbeat analysis data of the single-lead device, and there is no voting decision process. The above-mentioned classification methods are described separately below.
单导联分类方法包括：Single-lead classification methods include:
根据心搏时间序列数据，将单导联心搏数据进行切割生成单导联的心搏分析数据，并输入到训练得到的对应该导联的心搏分类模型进行幅值和时间表征数据的特征提取和分析，得到单导联的一次分类信息。According to the heartbeat time series data, the single-lead heartbeat data is cut to generate single-lead heartbeat analysis data, and input to the trained heartbeat classification model corresponding to the characteristics of the amplitude and time representation data Extract and analyze to obtain primary classification information of a single lead.
导联投票决策分类方法可以具体包括：Lead voting decision classification methods can specifically include:
第一步、根据心搏时间序列数据，对各导联心搏数据进行切割，从而生成各导联的心搏分析数据；The first step is to cut the heartbeat data of each lead according to the heartbeat time series data, so as to generate the heartbeat analysis data of each lead;
第二步、根据训练得到的各导联对应的心搏分类模型对各导联的心搏分析数据进行幅值和时间表征数据的特征提取和分析，得到各导联的分类信息；In the second step, the heartbeat analysis data of each lead is extracted and analyzed according to the heartbeat classification model corresponding to each lead obtained through training, and the classification information of each lead is obtained;
第三步、根据各导联的分类信息和导联权重值参考系数进行分类投票决策计算，得到所述一次分类信息。具体的，导联权重值参考系数是基于心电数据贝叶斯统计分析得到各导联对不同心搏分类的投票权重系数。In the third step, the classification voting decision calculation is performed according to the classification information of each lead and the weight value reference coefficient of the lead to obtain the primary classification information. Specifically, the lead weight value reference coefficient is based on the Bayesian statistical analysis of the electrocardiographic data to obtain the voting weight coefficients of each lead for different heart beat classifications.
导联同步关联分类方法可以具体包括：The lead synchronization association classification method may specifically include:
根据心搏时间序列数据，对各导联心搏数据进行切割，从而生成各导联的心搏分析数据；然后根据训练得到的多导联同步关联分类模型对各导联的心搏分析数据进行同步幅值和时间表征数据的特征提取和分析，得到心搏分析数据的一次分类信息。According to the heartbeat time series data, cut the heartbeat data of each lead to generate the heartbeat analysis data of each lead; The feature extraction and analysis of the amplitude and time representation data are synchronized, and the primary classification information of the heartbeat analysis data is obtained.
心搏数据的同步关联分类方法输入是动态心电图设备所有导联数据，按照心搏分析数据统一的心搏位点，截取各导联上相同位置和一定长度的数据点，同步输送给经过训练的人工智能深度学习模型进行计算分析，输出是每个心搏位置点综合考虑了所有导联心电图信号特征，以及心搏在时间上前后关联的心律特征的准确心搏分类。Synchronous correlation classification method of heartbeat data The input is all the lead data of the dynamic electrocardiogram equipment. According to the unified heartbeat position of the heartbeat analysis data, the data points at the same position and a certain length on each lead are intercepted and sent synchronously to the trained The artificial intelligence deep learning model performs calculation and analysis, and the output is an accurate heartbeat classification that comprehensively considers the ECG signal characteristics of all leads and the heart rhythm characteristics related to the heartbeat in time for each heartbeat location point.
本方法充分考虑了心电图不同导联数据实际上就是测量了心脏电信号在不同的心电轴向量方向传递的信息流，把心电图信号在时间和空间上传递的多维度数字特征进行综合分析，极大地改进了传统方法仅仅依靠单个导联独立分析，然后把结果汇总进行一些统计学的投票方式而比较容易得出的分类错误的缺陷，极大地提高了心搏分类的准确率。This method fully considers that the different lead data of the electrocardiogram is actually the measurement of the information flow of the cardiac electrical signal transmitted in different vector directions of the electrocardiogram, and comprehensively analyzes the multi-dimensional digital characteristics of the electrocardiogram signal transmitted in time and space. It has greatly improved the defect that the traditional method only relies on the independent analysis of a single lead, and then aggregates the results for some statistical voting methods, which is relatively easy to obtain classification errors, and greatly improves the accuracy of heart beat classification.
本步骤中采用的心搏分类模型可以如图4所示，具体可以为基于人工智能深度学习的卷积神经网络AlexNet，VGG16，Inception等模型启发的端对端多标签分类模型。具体的讲，该模型的网络是一个7层的卷积网络，每个卷积之后紧跟一个激活函数。第一层是两个不同尺度的卷积层，之后是六个卷积层。七层卷积的卷积核分别是96,256,256,384,384,384,256。除第一层卷积核有两个尺度分别是5和11外，其他层卷积核尺度为5。第三、五、六、七层卷积层后是池化层。最后跟着两个全连接层。The heart beat classification model used in this step can be shown in Figure 4, specifically, it can be an end-to-end multi-label classification model inspired by artificial intelligence deep learning-based convolutional neural network AlexNet, VGG16, Inception and other models. Specifically, the network of the model is a 7-layer convolutional network, and each convolution is followed by an activation function. The first layer is two convolutional layers of different scales, followed by six convolutional layers. The convolution kernels of the seven-layer convolution are 96, 256, 256, 384, 384, 384, 256, respectively. Except that the convolution kernel of the first layer has two scales of 5 and 11, the scale of the convolution kernel of other layers is 5. The third, fifth, sixth, and seventh convolutional layers are followed by pooling layers. Finally followed by two fully connected layers.
本步骤中的心搏分类模型，我们采用了训练集包含30万病人的1700万数据样本进行训练。这些样本是根据动态心电图分析诊断的要求对数据进行准确的标注产生的，标注主要是针对常见心律失常，传导阻滞以及ST段和T波改变，可满足不同应用场景的模型训练。具体以预设标准数据格式保存标注的信息。在训练数据的预处理上，为增加模型的泛化能力，对于样本量较少的分类做了小幅的滑动来扩增数据，具体的说，就是以每个心搏为基础，按照一定步长(比如10-50个数据点)移动2次，这样就可以增加2倍的数据，提高了对这些数据量比较少的分类样本的识别准确率。经过实际结果验证，泛化能力也得到了改善。For the heart beat classification model in this step, we used a training set containing 17 million data samples of 300,000 patients for training. These samples are generated by accurately labeling the data according to the requirements of dynamic electrocardiogram analysis and diagnosis. The labeling is mainly for common arrhythmias, conduction blocks, and ST segment and T wave changes, which can meet the model training of different application scenarios. Specifically, the marked information is saved in a preset standard data format. In the preprocessing of the training data, in order to increase the generalization ability of the model, a small slide is made for the classification with a small sample size to amplify the data. Specifically, it is based on each heartbeat and according to a certain step size (For example, 10-50 data points) move twice, so that the data can be increased by 2 times, and the recognition accuracy of these classification samples with relatively small amount of data is improved. Verified by actual results, the generalization ability has also been improved.
在一个实际训练过程使用了两台GPU服务器进行几十次轮循训练，训练收敛后，使用500万独立的测试数据进行测试，准确率可以到达91.92％。In an actual training process, two GPU servers are used for dozens of round-robin training. After the training converges, 5 million independent test data are used for testing, and the accuracy rate can reach 91.92%.
其中，训练数据的截取的长度，可以是1秒到10秒。比如采样率是200Hz，以2.5s为采样长度，取得的数据长度是500个心电图电压值(毫伏)的一个片段D[500]，输入数据是：InputData(i，j),其中，i是第i个导联，j是导联i第j个片段D。输入数据全部经过随机打散才开始训练，保证了训练过程收敛，同时，控制从同一个病人的心电图数据中收集太多的样本，提高模型的泛化能力，既真实场景下的准确率。训练时候，同步输入所有导联对应的片段数据D，按照图像分析的多通道分析方法，对每个时间位置的多个空间维度(不同心电轴向量)的导联数据进行同步学习，从而得到一个比常规算法更准确的分类结果。Wherein, the interception length of the training data may be 1 second to 10 seconds. For example, the sampling rate is 200Hz, and the sampling length is 2.5s. The obtained data length is a segment D[500] of 500 ECG voltage values (millivolts). The input data is: InputData(i, j), where i is The i-th lead, j is the j-th segment D of lead i. All input data are randomly scattered before training, which ensures the convergence of the training process. At the same time, it controls the collection of too many samples from the ECG data of the same patient, improving the generalization ability of the model, which is the accuracy rate in real scenarios. During training, the segment data D corresponding to all leads are input synchronously, and the lead data of multiple spatial dimensions (different ECG vectors) at each time position are synchronously learned according to the multi-channel analysis method of image analysis, so that A more accurate classification result than conventional algorithms is obtained.
步骤137，对一次分类信息结果中的特定心搏的心搏分析数据输入到ST段和T波改变模型进行识别，确定ST段和T波评价信息；Step 137, input the heartbeat analysis data of a specific heartbeat in a classification information result into the ST segment and T wave change model for identification, and determine the ST segment and T wave evaluation information;
ST段和T波评价信息具体为心搏分析数据对应的ST段和T波发生改变的导联位置信息。因为临床诊断要求对于ST段和T波的改变定位到具体的导联。The ST-segment and T-wave evaluation information specifically refers to position information of leads where the ST-segment and T-wave corresponding to the heartbeat analysis data change. Because clinical diagnosis requires that changes in the ST segment and T waves be located in specific leads.
其中，一次分类信息的特定心搏数据是指包含窦性心搏(N)和其它可能包含ST改变的心搏类型的心搏分析数据。Wherein, the specific heartbeat data of one classification information refers to heartbeat analysis data including sinus heartbeat (N) and other heartbeat types that may include ST changes.
ST段和T波改变导联定位模块将一次分类信息的特定心搏数据，按照每个导联依次输入到一个为识别ST段和T波改变的人工智能深度学习训练模型，进行计算分析，输出的结果说明导联片段的特征是否符合ST段和T波改变的结论，这样就可以确定ST段和T波改变发生的在具体那些导联的信息，即ST段和T波评价信息。具体方法可以是：把一次分类信息中结果是窦性心搏的各导联心搏分析数据，输入给ST段和T波改变模型，对窦性心搏分析数据进行逐一识别判断，以确定窦性心搏分析数据是否存在ST段和T波改变特征以及发生的具体导联位置信息，确定ST段和T波评价信息。The ST segment and T wave change lead positioning module will input the specific heartbeat data of a classification information into an artificial intelligence deep learning training model for identifying ST segment and T wave changes according to each lead, perform calculation and analysis, and output The results show whether the characteristics of the lead segment conform to the conclusion of ST segment and T wave changes, so that the information of specific leads where ST segment and T wave changes occur can be determined, that is, ST segment and T wave evaluation information. The specific method can be: input the heartbeat analysis data of each lead in the classified information to the ST segment and T wave change model, and identify and judge the sinus heartbeat analysis data one by one to determine whether the sinus heartbeat analysis data is There are ST segment and T wave change characteristics and the specific lead position information where it occurs, and the ST segment and T wave evaluation information is determined.
本步骤中采用的ST段和T波改变模型可以如图5所示，具体可以为基于人工智能深度学习的卷积神经网络AlexNet和VGG16等模型启发的端对端分类模型。具体的讲，该模型是一个7层的网络，模型包含了7个卷积，5个池化和2个全连接。卷积使用的卷积核均为1x5，每层卷积的滤波器个数各不相同。第1层卷积滤波器个数为96；第2层卷积和第3层卷积连用，滤波器个数为256；第4层卷积和第5层卷积连用，滤波器个数为384；第6层卷积滤波器个数为384；第7层卷积滤波器个数为256；第1、3、5、6、7层卷积层后是池化。随后是两个全连接，最后还采用Softmax分类器将结果分为两类。为了增加模型的非线性，提取数据更高维度的特征，故采用两个卷积连用的模式。The ST segment and T wave change model used in this step can be shown in Figure 5, specifically, it can be an end-to-end classification model inspired by models such as convolutional neural network AlexNet and VGG16 based on artificial intelligence deep learning. Specifically, the model is a 7-layer network, which includes 7 convolutions, 5 pools and 2 full connections. The convolution kernels used for convolution are all 1x5, and the number of filters for each layer of convolution is different. The number of convolution filters in the first layer is 96; the number of filters in the second layer convolution and the third layer convolution is 256; the fourth layer convolution and the fifth layer convolution are used in conjunction, and the number of filters is 384; the number of convolutional filters in the 6th layer is 384; the number of convolutional filters in the 7th layer is 256; the 1st, 3rd, 5th, 6th, and 7th convolutional layers are followed by pooling. This is followed by two full connections, and finally a Softmax classifier is used to divide the results into two categories. In order to increase the nonlinearity of the model and extract higher-dimensional features of the data, two convolution modes are used in conjunction.
因为带有ST段和T波改变的心搏在所有心搏中的占比较低，为了兼顾训练数据的多样性及各个类别数据量的均衡性，选取无ST段和T波改变以及有ST段和T波改变的训练数据比例约为2:1，保证了模型在分类过程中良好的泛化能力且不出现对训练数据占比较多一类的倾向性。由于心搏的形态多种多样，不同个体表现的形态不尽相同，因此，为了模型更好估计各分类的分布，能有效提取特征，训练样本从不同年龄，体重，性别和居住地区的个体收集；另外，因为单个个体在同一时间段内的心电图数据往往是高度相似的，所以为了避免过度学习，在获取单个个体的数据时，从所有数据中随机选取不同时间段的少量样本；最后，由于患者的心搏形态存在个体间差异大，而个体内相似度高的特点，因而在划分训练、测试集时，把不同的患者分到不同的数据集，避免同一个体的数据同时出现在训练集与测试集中，由此，所得模型测试结果最接近真实应用场景，保证了模型的可靠性和普适性。Because heartbeats with ST segment and T wave changes account for a relatively low proportion of all heartbeats, in order to take into account the diversity of training data and the balance of data volume in each category, select no ST segment and T wave changes and ST segment The ratio of training data changed from T wave is about 2:1, which ensures the good generalization ability of the model in the classification process and does not appear to have a tendency to account for more training data. Due to the variety of forms of heartbeat, the forms of different individuals are different. Therefore, in order to better estimate the distribution of each category and extract features effectively, the training samples are collected from individuals of different ages, weights, genders and living areas. ; In addition, because the ECG data of a single individual in the same time period are often highly similar, so in order to avoid over-learning, when obtaining the data of a single individual, a small number of samples in different time periods are randomly selected from all the data; finally, due to The patient's heartbeat pattern has the characteristics of large inter-individual differences and high intra-individual similarity. Therefore, when dividing the training and test sets, different patients are divided into different data sets to avoid the data of the same individual appearing in the training set at the same time. In this way, the obtained model test results are closest to the real application scenarios, which ensures the reliability and universality of the model.
步骤138，根据心搏时间序列数据，对心搏分析数据进行P波和T波特征检测，确定每个心搏中P波和T波的详细特征信息；Step 138, according to the heartbeat time series data, perform P wave and T wave feature detection on the heartbeat analysis data, and determine the detailed characteristic information of the P wave and T wave in each heartbeat;
具体的，详细特征信息包括幅值、方向、形态和起止时间的数据；在对心搏信号的分析中，P波、T波以及QRS波中的各项特征也是心电图分析中的重要依据。Specifically, the detailed feature information includes amplitude, direction, shape, and start-stop time data; in the analysis of heartbeat signals, various features in P wave, T wave, and QRS wave are also important basis for ECG analysis.
在P波和T波特征检测模块中，通过计算QRS波群中切分点位置，以及P波和T波的切分点位置，来提取P波、T波以及QRS波群中的各项特征。可以分别通过QRS波群切分点检测、单导联PT检测算法和多导联PT检测算法来实现。In the P wave and T wave feature detection module, the features in the P wave, T wave and QRS wave group are extracted by calculating the position of the cut point in the QRS wave group and the cut point position of the P wave and T wave . It can be realized through QRS complex segmentation point detection, single-lead PT detection algorithm and multi-lead PT detection algorithm respectively.
QRS波群切分点检测：根据QRS波群检测算法提供的QRS波群段功率最大点以及起止点，寻找单个导联中QRS波群的R点，R’点，S点以及S’点。在存在多导联数据时，计算各个切分点的中位数作为最后的切分点位置。QRS complex segmentation point detection: According to the maximum power point and start and end points of the QRS complex segment provided by the QRS complex detection algorithm, find the R point, R’ point, S point and S’ point of the QRS complex in a single lead. When there is multi-lead data, the median of each cut point is calculated as the final cut point position.
单导联P波、T波检测算法：P波和T波相对QRS波群幅度低、信号平缓，容易淹没在低频噪声中，是检测中的难点。本方法依据QRS波群检测的结果，在消除QRS波群对低频频段的影响后，使用低通滤波器对信号进行第三滤波，使PT波相对幅度增高。之后通过峰值检测的方法在两个QRS波群之间寻找T波。因为T波是心室复极产生的波群，因此T波和QRS波群之间有明确的锁时关系。以检测到的QRS波群为基准，在每个QRS波群到下一个QRS波群间期取中点(比如限制在第一个QRS波群后400ms到600ms之间的范围)作为T波检测结束点，在此区间内选取最大的峰作为T波。再在剩余的峰值内选择幅度最大的峰为P波。同时也根据P波和T波的峰值与位置数据，确定P波和T波的方向和形态特征。优选的，低通滤波的截止频率设置为10-30Hz之间。Single-lead P wave and T wave detection algorithm: Compared with the QRS complex, the amplitude of P wave and T wave is low, and the signal is flat, and it is easy to be submerged in low-frequency noise, which is a difficult point in detection. Based on the results of QRS complex detection, this method uses a low-pass filter to perform a third filter on the signal after eliminating the influence of the QRS complex on the low-frequency band to increase the relative amplitude of the PT wave. The T wave is then searched between the two QRS complexes by means of peak detection. Because the T wave is the complex generated by ventricular repolarization, there is a clear time-locked relationship between the T wave and the QRS complex. Based on the detected QRS complex, take the midpoint between each QRS complex and the next QRS complex (such as limited to the range between 400ms and 600ms after the first QRS complex) as T wave detection The end point, select the largest peak in this interval as the T wave. Then select the peak with the largest amplitude as the P wave among the remaining peaks. At the same time, according to the peak and position data of P wave and T wave, the direction and shape characteristics of P wave and T wave are determined. Preferably, the cutoff frequency of the low-pass filter is set between 10-30 Hz.
多导联P波、T波检测算法：在多导联的情况中，由于心搏中各个波的产生时间相同，空间分布不同，而噪声的时间空间分布不同，可以通过溯源算法来进行P、T波的检测。首先对信号进行QRS波群消除处理并使用低通滤波器对信号进行第三滤波以去除干扰。之后通过独立成分分析算法计算原始波形中的各个独立成分。在分离出的各个独立成分中，依据其峰值的分布特征以及QRS波群位置，选取相应的成分作为P波和T波信号，同时确定P波和T波的方向和形态特征。Multi-lead P wave and T wave detection algorithm: In the case of multi-lead, since the generation time of each wave in the heartbeat is the same, the spatial distribution is different, and the temporal and spatial distribution of the noise is different, the P, T wave can be detected by the traceability algorithm. Detection of T waves. Firstly, QRS complex elimination processing is performed on the signal and a third filtering is performed on the signal using a low-pass filter to remove interference. Each independent component in the original waveform is then calculated by an independent component analysis algorithm. Among the separated independent components, according to the distribution characteristics of their peaks and the position of the QRS complex, the corresponding components are selected as the P wave and T wave signals, and the direction and shape characteristics of the P wave and T wave are determined at the same time.
步骤139，对心搏分析数据在一次分类信息下根据心电图基本规律参考数据、P波和T波的详细特征信息以及ST段和T波评价信息进行二次分类处理，得到心搏分类信息；对心搏分类信息进行分析匹配，生成心电图事件数据。Step 139, performing secondary classification processing on the heartbeat analysis data under the primary classification information according to the basic law reference data of the electrocardiogram, detailed feature information of P waves and T waves, and ST segment and T wave evaluation information, to obtain heartbeat classification information; Heart beat classification information is analyzed and matched to generate ECG event data.
具体的，心电图基本规律参考数据是遵循权威心电图教科书中对心肌细胞电生理活动和心电图临床诊断的基本规则描述生成的，比如两个心搏之间最小的时间间隔，P波与R波的最小间隔等等，用于将心搏分类后的一次分类信息再进行细分；主要根据是心搏间RR间期以及不同心搏信号在各导联上的医学显著性；心搏审核模块依据心电图基本规律参考数据结合一定连续多个心搏分析数据的分类识别，以及P波和T波的详细特征信息将室性心搏分类拆分更细的心搏分类，包括：室性早搏(V)、室性逸搏(VE)、加速性室性早搏(VT)，将室上性类心搏细分为室上性早搏(S)、房性逸搏(SE)、交界性逸搏(JE)和房性加速性早搏(AT)等等。Specifically, the reference data of the basic laws of the electrocardiogram are generated following the description of the basic rules of the electrophysiological activity of cardiomyocytes and the clinical diagnosis of electrocardiograms in authoritative electrocardiogram textbooks, such as the minimum time interval between two heartbeats, the minimum time interval between P waves and R waves Interval, etc., are used to subdivide the primary classification information after heartbeat classification; the main basis is the RR interval between heartbeats and the medical significance of different heartbeat signals on each lead; the heartbeat review module is based on the electrocardiogram The basic law reference data combined with the classification and identification of certain continuous multiple heartbeat analysis data, as well as the detailed characteristic information of P wave and T wave, split the ventricular heartbeat classification into more detailed heartbeat classification, including: premature ventricular beat (V) , Ventricular escape (VE), accelerated ventricular premature beat (VT), the supraventricular beat is subdivided into supraventricular premature beat (S), atrial escape (SE), junctional escape (JE ) and atrial accelerated premature beats (AT) and so on.
此外，通过二次分类处理，还可以纠正一次分类中发生的不符合心电图基本规律参考数据的错误分类识别。将细分后的心搏分类按照心电图基本规律参考数据进行模式匹配，找到不符合心电图基本规律参考数据的分类识别，根据RR间期及前后分类标识纠正为合理的分类。In addition, through the secondary classification process, it is also possible to correct the misclassification and recognition that occurred in the primary classification that does not conform to the reference data of the basic rules of the electrocardiogram. The subdivided heart beat classification is pattern-matched according to the basic law reference data of the electrocardiogram, and the classification identification that does not conform to the basic law reference data of the electrocardiogram is found, and the reasonable classification is corrected according to the RR interval and the classification marks before and after.
具体的，经过二次分类处理，可以输出多种心搏分类，比如：正常窦性心搏(N)、完全性右束支阻滞(N_CRB)、完全性左束支阻滞(N_CLB)、室内阻滞(N_VB)、一度房室传导阻滞(N_B1)、预激(N_PS)、室性早搏(V)、室性逸搏(VE)、加速性室性早搏(VT)、室上性早搏(S)、房性逸搏(SE)、交界性逸搏(JE)、加速性房性早搏(AT)、房扑房颤(AF)、伪差(A)等分类结果。Specifically, after secondary classification processing, various heartbeat classifications can be output, such as: normal sinus beat (N), complete right bundle branch block (N_CRB), complete left bundle branch block (N_CLB), intraventricular block Block (N_VB), first-degree atrioventricular block (N_B1), pre-excitation (N_PS), ventricular premature beat (V), ventricular escape (VE), accelerated ventricular premature beat (VT), supraventricular premature beat ( S), atrial escape (SE), junctional escape (JE), accelerated atrial premature beat (AT), atrial flutter and atrial fibrillation (AF), artifact (A) and other classification results.
通过本步骤，还可以完成基础心率参数的计算。其中基础计算的心率参数包括：RR间期、心率、QT时间、QTc时间等参数。Through this step, the calculation of the basic heart rate parameters can also be completed. The heart rate parameters for basic calculation include: RR interval, heart rate, QT time, QTc time and other parameters.
随后，根据心搏二次分类结果，按照心电图基本规律参考数据进行模式匹配，可以得到分类对应于心电图事件数据的以下这些典型的心电图事件，包括但不限于：Subsequently, according to the result of the secondary classification of the heart beat, pattern matching is performed according to the reference data of the basic rules of the electrocardiogram, and the following typical electrocardiogram events corresponding to the classification of the electrocardiogram event data can be obtained, including but not limited to:
室上性早搏supraventricular premature beat
室上性早搏成对Paired supraventricular premature beats
室上性早搏二联律supraventricular premature beat bigeminy
室上性早搏三联律supraventricular premature beat triad
房性逸搏atrial escape
房性逸搏心律atrial escape rhythm
交界性逸搏junctional escape
交界性逸搏心律junctional escape rhythm
非阵发性室上性心动过速nonparoxysmal supraventricular tachycardia
最快室上性心动过速fastest supraventricular tachycardia
最长室上性心动过速longest supraventricular tachycardia
室上性心动过速supraventricular tachycardia
短阵室上性心动过速short supraventricular tachycardia
心房扑动-心房颤动Atrial Flutter - Atrial Fibrillation
室性早搏premature ventricular beats
室性早搏成对paired ventricular premature beats
室性早搏二联律premature ventricular contraction bigeminy
室性早搏三联律premature ventricular triad
室性逸搏Ventricular escape
室性逸搏心律ventricular escape rhythm
加速性室性自主心律accelerated ventricular spontaneous rhythm
最快室性心动过速fastest ventricular tachycardia
最长室性心动过速longest ventricular tachycardia
室性心动过速ventricular tachycardia
短阵室性心动过速short burst ventricular tachycardia
二度I型窦房传导阻滞Second degree type I sinoatrial block
二度II型窦房传导阻滞Second degree type II sinoatrial block
一度房室传导阻滞first degree atrioventricular block
二度I型房室传导阻滞Second degree type I atrioventricular block
二度II型房室传导阻滞Second degree type II atrioventricular block
二度II型(2：1)房室传导阻滞Second degree type II (2:1) atrioventricular block
高度房室传导阻滞high degree atrioventricular block
完全性左束支阻滞complete left bundle branch block
完全性右束支阻滞complete right bundle branch block
室内阻滞indoor block
预激综合症pre-excitation syndrome
ST段和T波改变ST segment and T wave changes
最长RR间期longest RR interval
将心搏分析数据根据心搏分类信息和心电图基本规律参考数据生成心电图事件数据。心电事件数据包括动态监护设备的设备ID信息。The heartbeat analysis data is generated according to the heartbeat classification information and the basic rule reference data of the electrocardiogram to generate electrocardiogram event data. The ECG event data includes device ID information of the dynamic monitoring device.
步骤140，根据心电图事件数据实时确定对应的心电图事件信息，并确定心电图事件信息是否为心电异常事件信息；Step 140, determine the corresponding ECG event information in real time according to the ECG event data, and determine whether the ECG event information is abnormal ECG event information;
具体的，在得到心电图事件数据之后，可以通过人工智能学习得到的心电图事件数据与心电图时间信息的对应关系，对应得到相应的心电事件信息，比如，心电图事件数据对应的心电事件信息为窦性心搏事件、室性早搏事件等。这其中仅有部分为需要产生报警的心电异常事件。Specifically, after obtaining the electrocardiogram event data, the corresponding relationship between the electrocardiogram event data and the electrocardiogram time information obtained through artificial intelligence learning can be used to obtain the corresponding electrocardiogram event information. For example, the electrocardiogram event data corresponding to the electrocardiogram event information is sinus rhythm stroke events, ventricular premature beats, etc. Only some of them are abnormal ECG events that need to generate an alarm.
上述的数据处理过程均为实时的，因此动态监护设备会不断有心电图事件信息的产生。在实际应用中也可以合理设定心电图事件信息的输出间隔，既减小数据运算量，又避免漏检的情况。The above-mentioned data processing processes are all real-time, so the dynamic monitoring equipment will continuously generate electrocardiogram event information. In practical applications, the output interval of the electrocardiogram event information can also be reasonably set, which not only reduces the amount of data calculation, but also avoids the situation of missed detection.
在得到心电图事件信息后，与动态监护设备中记录的心电异常事件信息进行匹配，当为心电异常事件信息时，执行步骤150，否则继续执行步骤120，对被测对象继续进行监护数据采集。After the electrocardiogram event information is obtained, it is matched with the abnormal ECG event information recorded in the dynamic monitoring device. When it is the abnormal ECG event information, step 150 is performed, otherwise step 120 is continued to continue monitoring data collection for the measured object .
在优选的方案中，无论是否监测到发生心电异常事件，根据心电图数据、心电图事件数据可以按照预设规则生成针对被监测用户的监控报告。In a preferred solution, no matter whether an abnormal ECG event is detected or not, a monitoring report for the monitored user can be generated according to preset rules based on the ECG data and the ECG event data.
比如，可以根据预设时间间隔，例如每24小时汇总一次数据，生成相应时段内被测对象的监测报告数据。For example, the data can be summarized according to a preset time interval, for example, every 24 hours, and the monitoring report data of the measured object in the corresponding period can be generated.
步骤150，生成并通过动态监护设备输出报警信息；Step 150, generate and output alarm information through dynamic monitoring equipment;
具体的，当确定心电图事件信息为心电异常事件信息时，如果前述对心电图数据进行分析处理的过程是在动态监护设备中执行的，则通过动态监护设备直接输出根据心电异常事件信息产生相应的报警信息。Specifically, when it is determined that the electrocardiogram event information is the abnormal electrocardiogram event information, if the aforementioned process of analyzing and processing the electrocardiogram data is performed in the dynamic monitoring device, the dynamic monitoring device will directly output the corresponding alarm information.
如果前述对心电图数据进行分析处理的过程是在服务器中执行的，则服务器根据心电异常事件信息产生相应的报警信息，并根据设备ID将报警信息实时下发到动态监护设备，通过动态监护设备输出。当然，也可以将心电异常事件信息同步发送，以通过动态监护设备输出更多的信息。If the above-mentioned process of analyzing and processing the electrocardiogram data is performed in the server, the server generates corresponding alarm information according to the abnormal event information of the electrocardiogram, and sends the alarm information to the dynamic monitoring device in real time according to the device ID. output. Of course, the abnormal ECG event information can also be sent synchronously, so as to output more information through the dynamic monitoring device.
报警信息的输出，在于提示被监测者其监控数据发生异常，使得被监测者及时了解自身状况，同时在被监测者附近的其他人员也能及时了解被监测者发生了监测数据异常，从而实现实时的异常告警。The output of the alarm information is to remind the monitored person that his monitoring data is abnormal, so that the monitored person can understand his own situation in time, and other personnel near the monitored person can also know in time that the monitored person has abnormal monitoring data, so as to realize real-time monitoring. abnormal alarm.
更为优选的，报警信息包括：设备ID信息、被测对象信息和心电异常事件信息，从而得以形成完整的信息输出。More preferably, the alarm information includes: device ID information, measured object information and abnormal ECG event information, so as to form a complete information output.
在同时考虑其他体征监护数据时，可以在脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据中的一个或多个存在超出相应的设定阈值的异常数据时，并结合心电图事件数据生成体征监测异常事件信息，来确定报警信息的输出。When considering other sign monitoring data at the same time, when one or more of the pulse data, blood pressure data, respiration data, blood oxygen saturation data and body temperature data has abnormal data exceeding the corresponding set threshold, it can be combined with ECG events The data generates signs and monitors abnormal event information to determine the output of alarm information.
步骤160，获取动态监护设备的工作模式；Step 160, acquiring the working mode of the dynamic monitoring device;
在具体的应用场景中，本发明的动态监护设备可以被配置为不同的工作模式，即根据不同的工作模式提供不同的服务。工作模式至少可以包括本地处理模式和后台处理模式。In specific application scenarios, the dynamic monitoring device of the present invention can be configured in different working modes, that is, different services are provided according to different working modes. The working modes may at least include a local processing mode and a background processing mode.
在本地处理模式下，异常事件信息和相关数据仅在本地存储，不上报到后台处理。In the local processing mode, abnormal event information and related data are only stored locally and are not reported to the background for processing.
而在后台处理模式下，异常事件信息和相关数据被上报到服务器，并进行分派处理，使得被测对象的异常情况能够得到及时有效的处置，得到更好的医疗救助服务。In the background processing mode, abnormal event information and related data are reported to the server and dispatched for processing, so that the abnormal situation of the measured object can be dealt with in a timely and effective manner, and better medical assistance services can be obtained.
配置可以在动态监护设备本地完成，也可以在动态监护设备被分配给用户使用时，由后台操作完成。当然，选用不同的工作模式即不同的服务，其收费方式也可以相应不同。The configuration can be completed locally on the dynamic monitoring device, or it can be completed by background operation when the dynamic monitoring device is assigned to the user. Of course, different working modes, that is, different services are selected, and the charging methods may also be correspondingly different.
步骤170，当为本地处理模式时，动态监护设备对心电图事件数据和报警信息进行记录存储；Step 170, when it is in the local processing mode, the dynamic monitoring device records and stores the electrocardiogram event data and alarm information;
步骤180，当为后台处理模式时，服务器根据报警信息或根据接收到动态监护设备基于报警信息触发的报警信号，确定被测对象信息对应的责任用户的用户ID，并生成通知信息发送给责任用户的用户设备和/或预设的关联机构的用户设备；Step 180, when it is in the background processing mode, the server determines the user ID of the responsible user corresponding to the measured object information according to the alarm information or according to the received alarm signal triggered by the dynamic monitoring device based on the alarm information, and generates notification information to send to the responsible user and/or the user equipment of the preset associated organization;
具体的，如果是动态监护设备产生报警信息，则动态监护设备可以将报警信息发送给服务器，或者，动态监护设备向服务器发送报警信号，以告知服务器有报警信息产生，当然在报警信号中携带有设备ID信息和被测对象信息。此时，服务器根据报警信息或者报警信号中携带的被测对象信息，确定与被测对象相关联的责任用户或关联机构的信息。Specifically, if the dynamic monitoring device generates alarm information, the dynamic monitoring device can send the alarm information to the server, or the dynamic monitoring device sends an alarm signal to the server to inform the server that an alarm message is generated. Of course, the alarm signal carries Device ID information and measured object information. At this time, the server determines the information of the responsible user or associated institution associated with the measured object according to the alarm information or the measured object information carried in the alarm signal.
如果报警信息是在服务器生成的，则服务器根据报警信息中的被测对象信息就可确定与被测对象相关联的责任用户或关联机构的信息。If the alarm information is generated on the server, the server can determine the information of the responsible user or associated institution associated with the measured object according to the measured object information in the alarm information.
这里需要说明的是，在服务器中预先存储有与被监测用户相关联的责任用户或预设的关联机构的信息。责任用户可以是被监测用户的监护人、亲属、家庭医生等，责任用户的用户设备可以是上述人群的安装并运行有相应应用的智能手机、平板电脑或者其他至少具有信息接收和显示功能的设备；关联机构可以是被监测用户指定的医疗机构等，同样的，关联结构的用户设备亦可以是安装并运行有相应应用的智能手机、平板电脑或者其他至少具有信息接收和显示功能的设备。It should be noted here that the server pre-stores the information of the responsible user or the preset associated organization associated with the monitored user. The responsible user can be the guardian, relative, family doctor, etc. of the monitored user, and the user equipment of the responsible user can be a smart phone, a tablet computer or other devices with at least information receiving and display functions installed and running corresponding applications of the above-mentioned people; The associated institution may be a medical institution designated by the monitored user, and similarly, the user device of the associated structure may also be a smart phone, a tablet computer or other devices with at least information receiving and display functions installed and running corresponding applications.
服务器根据设备ID信息确定动态监护设备的位置信息，从而确定被监测对象所处位置。The server determines the location information of the dynamic monitoring device according to the device ID information, thereby determining the location of the monitored object.
服务器根据报警信息或者报警信号，生成通知信息，在通知信息中至少携带有被测对象信息，以及根据设备ID确定的被监测对象所处位置的位置信息，使得接收者即责任用户至少能够得知发生异常报警的对象，从而能够快速与该被监测者进行联系。在具体应用中，被测对象信息可以包括被测对象的手机号码等，以方便联系被测对象，向被测对象提供帮助。The server generates notification information according to the alarm information or alarm signal, and the notification information carries at least the information of the measured object and the location information of the monitored object determined according to the device ID, so that the recipient, that is, the responsible user, can at least know The object that has an abnormal alarm, so that the monitored person can be quickly contacted. In a specific application, the measured object information may include the measured object's mobile phone number, etc., so as to facilitate contact with the measured object and provide help to the measured object.
在优选的方案中，通知信息中还包括有心电异常事件信息，使得通知信息的接收者能够根据心电异常时间信息进行预判断，从而在第一时间内就能够对异常情况进行了解，可以根据其严重程度采取相应的医疗救助措施。其中包括向动态监护设备发送反馈信息，提示被监测者发生异常及临时应对措施，比如服用药物、静坐、迅速就医、等待医疗机构上门服务等。In a preferred solution, the notification information also includes abnormal ECG event information, so that the recipient of the notification information can make a pre-judgment based on the abnormal ECG time information, so that the abnormal situation can be understood in the first time, and can be based on Take appropriate medical assistance measures according to its severity. These include sending feedback information to the dynamic monitoring equipment, prompting the monitored person to experience abnormalities and temporary countermeasures, such as taking drugs, sitting still, seeking medical treatment quickly, and waiting for on-site services from medical institutions.
具体的，在步骤160之后，服务器接收责任用户的用户设备发送的反馈信息，并根据设备ID信息将反馈信息发送给动态监护设备；由动态监护设备输出反馈信息。Specifically, after step 160, the server receives the feedback information sent by the user equipment of the responsible user, and sends the feedback information to the dynamic monitoring device according to the device ID information; the dynamic monitoring device outputs the feedback information.
上述对本发明实施例的体征信息动态监护方法的执行过程进行了详述，为了便于理解，下面我们以实际应用的一个具体例子为例进行说明。The execution process of the method for dynamic monitoring of vital sign information according to the embodiment of the present invention has been described above in detail. For ease of understanding, a specific example of practical application will be used as an example below for illustration.
在实际应用，我们可以在本地处理模式和后台处理模式的基础上进一步细分服务模式，这样可以基于同样的监护设备，提供用户更加定制化的监护服务。In practical application, we can further subdivide the service mode on the basis of local processing mode and background processing mode, so that users can be provided with more customized monitoring services based on the same monitoring equipment.
在一个具体的例子中，可以设定如下三种服务模式：In a specific example, the following three service modes can be set:
完整服务模式、紧急服务模式和本地服务模式。Full service mode, emergency service mode and local service mode.
在完整服务模式中，动态监护设备将监测得到的体征监护数据实时上传至服务器，并由服务器实时分发至关联机构的终端设备，实现24小时全天候的监控服务，并且，根据预设时间间隔，例如每24小时，汇总一次体征监护数据，生成相应时段内被测对象的监测报告数据。In the complete service mode, the dynamic monitoring equipment uploads the monitored physical signs monitoring data to the server in real time, and the server distributes it to the terminal equipment of the affiliated organization in real time, realizing 24-hour monitoring service, and, according to the preset time interval, such as Every 24 hours, the sign monitoring data is summarized, and the monitoring report data of the measured object in the corresponding period is generated.
当发生异常产生报警信息时，可以根据报警信息的严重程度采用不同的处理方式。When an abnormality occurs and alarm information is generated, different processing methods can be adopted according to the severity of the alarm information.
比如，如果发生的是一般严重程度的心电异常事件的报警，可以通过反馈信息的方式，进行远程指导服务；如果发生的是严重程度较高的报警，则通过服务器进行医疗救护任务派发，由相关医疗结构提供上门医疗服务，或者由相关医疗机构安排被监测者就医。For example, if an alarm of an abnormal ECG event of general severity occurs, remote guidance services can be provided through feedback information; if an alarm of high severity occurs, medical rescue tasks will be dispatched through the server, The relevant medical institutions provide door-to-door medical services, or the relevant medical institutions arrange for the monitored persons to seek medical treatment.
在紧急服务模式下，动态监护设备仅在产生报警信息时上报服务器，，同样的服务器可以根据报警信息的严重程度采用不同的处理方式。在没有发生异常报警时，动态监护设备仅在本地进行数据监控和显示。In the emergency service mode, the dynamic monitoring equipment only reports to the server when an alarm information is generated, and the same server can adopt different processing methods according to the severity of the alarm information. When there is no abnormal alarm, the dynamic monitoring equipment only monitors and displays data locally.
在本地服务模式下，动态监护设备只进行本地监控，在产生报警信息时也仅在本地输出。被监测者可以在得到报警信息后以手动触发发送报警信号的方式上报服务器。In the local service mode, the dynamic monitoring equipment only performs local monitoring, and only outputs locally when an alarm message is generated. The monitored person can report to the server by manually triggering and sending an alarm signal after receiving the alarm information.
此外，本发明的动态监护设备，还能够提供在用户自觉不适时的主动事件记录和上报的功能，以更好的为用户提供医疗救助服务和监控。In addition, the dynamic monitoring device of the present invention can also provide the function of active event recording and reporting when the user feels unwell, so as to better provide medical assistance services and monitoring for the user.
在动态监护设备上可以设置有一键触发的开关，当用户触发开关后，启动监听设备监听用户输入，并生成报警事件记录信息。监听设备可以包括但不限于：麦克风、摄像头、触摸屏、虚拟键盘等。用户可以通过视频、语音、文字输入等方式，进行自觉异常的症状描述。A one-button trigger switch can be set on the dynamic monitoring device. When the user triggers the switch, the monitoring device is started to monitor user input and generate alarm event record information. Monitoring devices may include, but are not limited to: microphones, cameras, touch screens, virtual keyboards, etc. Users can describe the symptoms of perceived abnormalities through video, voice, text input, etc.
进一步的，可以同时启动对用户触发开关前后的心电图数据以及脉搏数据、血压数据、呼吸数据、血氧饱和度数据和体温数据的截取，与用户输入的信息一并生成报警事件记录信息。Furthermore, the interception of ECG data before and after the user triggers the switch, pulse data, blood pressure data, respiration data, blood oxygen saturation data and body temperature data can be started simultaneously, and alarm event record information can be generated together with the information input by the user.
然后，将报警事件记录信息发送到后台服务器，并分发给责任用户和/或预设的关联机构的用户设备，以便及时进行处置。Then, the alarm event record information is sent to the background server, and distributed to the responsible user and/or the user equipment of the preset associated organization, so as to deal with it in time.
图6为本发明实施例提供的一种动态监护系统的结构示意图，该监护系统包括一个或多个动态监护设备和服务器。服务器及动态监护设备分别包括：处理器和存储器。存储器可通过总线与处理器连接。存储器可以是非易失存储器，例如硬盘驱动器和闪存，存储器中存储有软件程序和设备驱动程序。软件程序能够执行本发明实施例提供的上述方法的各种功能；设备驱动程序可以是网络和接口驱动程序。处理器用于执行软件程序，该软件程序被执行时，能够实现本发明实施例提供的方法。Fig. 6 is a schematic structural diagram of a dynamic monitoring system provided by an embodiment of the present invention, the monitoring system includes one or more dynamic monitoring devices and a server. The server and the dynamic monitoring device respectively include: a processor and a memory. The memory can be connected to the processor through the bus. The memory can be non-volatile memory, such as a hard drive and flash memory, where software programs and device drivers are stored. The software program can execute various functions of the above method provided by the embodiment of the present invention; the device driver can be a network and interface driver. The processor is configured to execute a software program. When the software program is executed, the method provided by the embodiment of the present invention can be implemented.
本发明实施例提供的面向用户的体征信息动态监护方法和动态监护系统，采用数据的预处理，心搏特征检测，基于深度学习方法的干扰信号检测和心搏分类与导联合并，心搏的审核，心电图事件和参数的分析计算，最终自动输出心电事件结果数据的一个完整快速流程的自动化心电检测分析，并且可基于心电检测分析结果输出报警，或结合血压、血氧、脉搏、呼吸、体温数据产生报警，以及基于报警的响应处理，通过基于报警信息进行信息分发处理，包括分发给医疗机构或者分发给被监测者的关联用户的终端设备，使得被监测者得到有效、及时的医疗救助服务。本发明的面向用户的体征信息动态监护方法和动态监护系统，面向非住院人群进行有效的体征监护，并基于体征监护为用户提供更有效的医疗保障服务。The user-oriented dynamic monitoring method and dynamic monitoring system for physical sign information provided by the embodiments of the present invention adopt data preprocessing, heartbeat feature detection, interference signal detection based on deep learning methods, heartbeat classification and lead combination, heartbeat Review, analysis and calculation of ECG events and parameters, and automatic ECG detection and analysis of a complete and fast process of automatically outputting ECG event result data, and can output alarms based on ECG detection and analysis results, or combine blood pressure, blood oxygen, pulse, Respiratory and body temperature data generate alarms, and alarm-based response processing, through information distribution processing based on alarm information, including distribution to medical institutions or terminal devices of associated users of the monitored person, so that the monitored person can get effective and timely Medical assistance service. The user-oriented dynamic monitoring method for sign information and the dynamic monitoring system of the present invention perform effective sign monitoring for non-hospitalized populations, and provide users with more effective medical security services based on the sign monitoring.
专业人员应该还可以进一步意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、计算机软件或者二者的结合来实现，为了清楚地说明硬件和软件的可互换性，在上述说明中已经按照功能一般性地描述了各示例的组成及步骤。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本发明的范围。Professionals should further realize that the units and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the relationship between hardware and software Interchangeability. In the above description, the composition and steps of each example have been generally described according to their functions. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present invention.
结合本文中所公开的实施例描述的方法或算法的步骤可以用硬件、处理器执行的软件模块，或者二者的结合来实施。软件模块可以置于随机存储器(RAM)、内存、只读存储器(ROM)、电可编程ROM、电可擦除可编程ROM、寄存器、硬盘、可移动磁盘、CD-ROM、或技术领域内所公知的任意其它形式的存储介质中。The steps of the methods or algorithms described in connection with the embodiments disclosed herein may be implemented by hardware, software modules executed by a processor, or a combination of both. Software modules can be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other Any other known storage medium.
以上所述的具体实施方式，对本发明的目的、技术方案和有益效果进行了进一步详细说明，所应理解的是，以上所述仅为本发明的具体实施方式而已，并不用于限定本发明的保护范围，凡在本发明的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。The specific embodiments described above have further described the purpose, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above descriptions are only specific embodiments of the present invention and are not intended to limit the scope of the present invention. Protection scope, within the spirit and principles of the present invention, any modification, equivalent replacement, improvement, etc., shall be included in the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810215889 | 2018-03-15 | ||
| CN2018102158893 | 2018-03-15 |
| Publication Number | Publication Date |
|---|---|
| CN108577830Atrue CN108577830A (en) | 2018-09-28 |
| CN108577830B CN108577830B (en) | 2021-10-26 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810222780.2AActiveCN108577830B (en) | 2018-03-15 | 2018-03-19 | User-oriented physical sign information dynamic monitoring method and system |
| Country | Link |
|---|---|
| CN (1) | CN108577830B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109497986A (en)* | 2018-11-22 | 2019-03-22 | 杭州脉流科技有限公司 | Electrocardiogram intelligent analysis method, device, computer equipment and system based on deep neural network |
| CN109620181A (en)* | 2019-01-16 | 2019-04-16 | 贝骨新材料科技（上海）有限公司 | A kind of method and nursing type custodial care facility of heart rate and heart rate variability monitoring |
| CN109875521A (en)* | 2019-04-18 | 2019-06-14 | 厦门纳龙科技有限公司 | A kind of analysis of ECG data and system |
| CN110598549A (en)* | 2019-08-07 | 2019-12-20 | 王满 | Convolutional neural network information processing system based on cardiac function monitoring and training method |
| CN110638460A (en)* | 2019-09-16 | 2020-01-03 | 深圳和而泰家居在线网络科技有限公司 | Method, device and equipment for detecting state of object relative to bed |
| CN112120693A (en)* | 2020-09-30 | 2020-12-25 | 毕胜普生物科技有限公司 | Electrocardio monitoring system and wearable equipment with same |
| CN112869753A (en)* | 2021-02-10 | 2021-06-01 | 上海交通大学医学院附属瑞金医院 | Analysis method, equipment, medium and electrocardiograph for QRST waveform of electrocardiogram |
| CN112911990A (en)* | 2018-12-24 | 2021-06-04 | 中国医学科学院北京协和医院 | Medical monitoring equipment and state information display method thereof |
| CN113110961A (en)* | 2021-04-30 | 2021-07-13 | 平安国际融资租赁有限公司 | Equipment abnormality detection method and device, computer equipment and readable storage medium |
| CN113509187A (en)* | 2021-07-12 | 2021-10-19 | 广州市康源图像智能研究院 | Real-time single-lead electrocardio R wave detection method |
| CN114072045A (en)* | 2019-10-29 | 2022-02-18 | 深圳迈瑞生物医疗电子股份有限公司 | Self-adaptive alarm system, method and device and Internet of things system |
| CN114120586A (en)* | 2020-08-25 | 2022-03-01 | 深圳迈瑞生物医疗电子股份有限公司 | Alarm method, monitoring device, monitoring system and readable storage medium |
| WO2022141012A1 (en)* | 2020-12-29 | 2022-07-07 | 深圳迈瑞生物医疗电子股份有限公司 | Medical monitoring manner setting method and related device |
| CN118526161A (en)* | 2024-06-17 | 2024-08-23 | 脑机交互与人机共融海河实验室 | In-vivo noninvasive brain function real-time monitoring system in hyperbaric oxygen chamber |
| CN118963553A (en)* | 2024-07-30 | 2024-11-15 | 济宁医学院附属医院 | A patient physiological information monitoring system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101036576A (en)* | 2007-04-29 | 2007-09-19 | 东北大学 | Portable pulse detecting device by continuous recoding |
| CN101363841A (en)* | 2007-08-09 | 2009-02-11 | 李秀 | Physiology situation information acquisition device |
| CN102138789A (en)* | 2011-01-24 | 2011-08-03 | 无锡微感科技有限公司 | Dynamic electrocardiogram and motion recording and analyzing system |
| CN102389303A (en)* | 2011-08-02 | 2012-03-28 | 长春华讯信息科技有限公司 | Method and device for monitoring electrocardiogram |
| CN104063645A (en)* | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
| CN104523266A (en)* | 2015-01-07 | 2015-04-22 | 河北大学 | Automatic classification method for electrocardiogram signals |
| CN104688219A (en)* | 2013-12-04 | 2015-06-10 | 乐普（北京）医疗器械股份有限公司 | Electrocardiosignal prediction method and electrocardiosignal prediction system |
| EP2944251A1 (en)* | 2014-05-13 | 2015-11-18 | Sensium Healthcare Limited | Method, apparatus and computer-readable medium for confidence level determination of ambulatory hr algorithm based on a three-way rhythm classifier |
| CN205433651U (en)* | 2016-02-23 | 2016-08-10 | 济宁中科大象医疗电子科技有限公司 | Wearable motion and electrocardiosignal are real -time to be gathered and guardianship equipment |
| US20160246937A1 (en)* | 2005-10-14 | 2016-08-25 | Medicalgorithmics S.A. | Systems for safe and remote outpatient ecg monitoring |
| CN106821366A (en)* | 2015-10-14 | 2017-06-13 | 张胜国 | Intelligent cardiac monitor system |
| US20170188967A1 (en)* | 2016-01-05 | 2017-07-06 | Tosense, Inc. | Physiological monitoring system featuring floormat and wired handheld sensor |
| WO2017172105A1 (en)* | 2016-03-29 | 2017-10-05 | Intel Corporation | Opportunistic measurements and processing of user's context |
| US20170367602A1 (en)* | 2014-07-07 | 2017-12-28 | Zoll Medical Corporation | System and Method for Distinguishing a Cardiac Event From Noise in an Electrocardiogram (ECG) Signal |
| CN107714023A (en)* | 2017-11-27 | 2018-02-23 | 乐普（北京）医疗器械股份有限公司 | Static ecg analysis method and apparatus based on artificial intelligence self study |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160246937A1 (en)* | 2005-10-14 | 2016-08-25 | Medicalgorithmics S.A. | Systems for safe and remote outpatient ecg monitoring |
| CN101036576A (en)* | 2007-04-29 | 2007-09-19 | 东北大学 | Portable pulse detecting device by continuous recoding |
| CN101363841A (en)* | 2007-08-09 | 2009-02-11 | 李秀 | Physiology situation information acquisition device |
| CN102138789A (en)* | 2011-01-24 | 2011-08-03 | 无锡微感科技有限公司 | Dynamic electrocardiogram and motion recording and analyzing system |
| CN102389303A (en)* | 2011-08-02 | 2012-03-28 | 长春华讯信息科技有限公司 | Method and device for monitoring electrocardiogram |
| CN104688219A (en)* | 2013-12-04 | 2015-06-10 | 乐普（北京）医疗器械股份有限公司 | Electrocardiosignal prediction method and electrocardiosignal prediction system |
| EP2944251A1 (en)* | 2014-05-13 | 2015-11-18 | Sensium Healthcare Limited | Method, apparatus and computer-readable medium for confidence level determination of ambulatory hr algorithm based on a three-way rhythm classifier |
| CN104063645A (en)* | 2014-07-01 | 2014-09-24 | 清华大学深圳研究生院 | Identity recognition method based on ECG dynamic self-updating samples |
| US20170367602A1 (en)* | 2014-07-07 | 2017-12-28 | Zoll Medical Corporation | System and Method for Distinguishing a Cardiac Event From Noise in an Electrocardiogram (ECG) Signal |
| CN104523266A (en)* | 2015-01-07 | 2015-04-22 | 河北大学 | Automatic classification method for electrocardiogram signals |
| CN106821366A (en)* | 2015-10-14 | 2017-06-13 | 张胜国 | Intelligent cardiac monitor system |
| US20170188967A1 (en)* | 2016-01-05 | 2017-07-06 | Tosense, Inc. | Physiological monitoring system featuring floormat and wired handheld sensor |
| CN205433651U (en)* | 2016-02-23 | 2016-08-10 | 济宁中科大象医疗电子科技有限公司 | Wearable motion and electrocardiosignal are real -time to be gathered and guardianship equipment |
| WO2017172105A1 (en)* | 2016-03-29 | 2017-10-05 | Intel Corporation | Opportunistic measurements and processing of user's context |
| CN107714023A (en)* | 2017-11-27 | 2018-02-23 | 乐普（北京）医疗器械股份有限公司 | Static ecg analysis method and apparatus based on artificial intelligence self study |
| Title |
|---|
| SAHOO S, 等: "Multiresolution wavelet transform based feature extraction and ECG classification to detect cardiac abnormalities", 《MEASUREMENT》* |
| 张飞: "异常心电波形的自动分析及其快速识别算法研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》* |
| 沈振宇,等: "实时动态心电监护系统", 《南京大学学报(自然科学版)》* |
| 陈义: "心电信号的异常心律分类算法研究", 《中国优秀硕士学位论文全文数据库 医药卫生科技辑》* |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109497986A (en)* | 2018-11-22 | 2019-03-22 | 杭州脉流科技有限公司 | Electrocardiogram intelligent analysis method, device, computer equipment and system based on deep neural network |
| CN112911990A (en)* | 2018-12-24 | 2021-06-04 | 中国医学科学院北京协和医院 | Medical monitoring equipment and state information display method thereof |
| CN109620181A (en)* | 2019-01-16 | 2019-04-16 | 贝骨新材料科技（上海）有限公司 | A kind of method and nursing type custodial care facility of heart rate and heart rate variability monitoring |
| CN109875521A (en)* | 2019-04-18 | 2019-06-14 | 厦门纳龙科技有限公司 | A kind of analysis of ECG data and system |
| CN110598549A (en)* | 2019-08-07 | 2019-12-20 | 王满 | Convolutional neural network information processing system based on cardiac function monitoring and training method |
| CN110638460A (en)* | 2019-09-16 | 2020-01-03 | 深圳和而泰家居在线网络科技有限公司 | Method, device and equipment for detecting state of object relative to bed |
| CN110638460B (en)* | 2019-09-16 | 2022-07-15 | 深圳数联天下智能科技有限公司 | Method, device and equipment for detecting state of object relative to bed |
| CN114072045A (en)* | 2019-10-29 | 2022-02-18 | 深圳迈瑞生物医疗电子股份有限公司 | Self-adaptive alarm system, method and device and Internet of things system |
| CN114072045B (en)* | 2019-10-29 | 2024-03-22 | 深圳迈瑞生物医疗电子股份有限公司 | Self-adaptive alarm system, method and device and Internet of things system |
| CN114120586A (en)* | 2020-08-25 | 2022-03-01 | 深圳迈瑞生物医疗电子股份有限公司 | Alarm method, monitoring device, monitoring system and readable storage medium |
| CN112120693A (en)* | 2020-09-30 | 2020-12-25 | 毕胜普生物科技有限公司 | Electrocardio monitoring system and wearable equipment with same |
| CN112120693B (en)* | 2020-09-30 | 2022-09-13 | 毕胜普生物科技有限公司 | Electrocardio monitoring system and wearable equipment with same |
| WO2022141012A1 (en)* | 2020-12-29 | 2022-07-07 | 深圳迈瑞生物医疗电子股份有限公司 | Medical monitoring manner setting method and related device |
| CN116669621A (en)* | 2020-12-29 | 2023-08-29 | 深圳迈瑞生物医疗电子股份有限公司 | Medical monitoring mode setting method and related equipment |
| CN112869753A (en)* | 2021-02-10 | 2021-06-01 | 上海交通大学医学院附属瑞金医院 | Analysis method, equipment, medium and electrocardiograph for QRST waveform of electrocardiogram |
| CN113110961A (en)* | 2021-04-30 | 2021-07-13 | 平安国际融资租赁有限公司 | Equipment abnormality detection method and device, computer equipment and readable storage medium |
| CN113509187A (en)* | 2021-07-12 | 2021-10-19 | 广州市康源图像智能研究院 | Real-time single-lead electrocardio R wave detection method |
| CN113509187B (en)* | 2021-07-12 | 2023-10-24 | 广州市康源图像智能研究院 | Real-time single-lead electrocardiograph R wave detection method |
| CN118526161A (en)* | 2024-06-17 | 2024-08-23 | 脑机交互与人机共融海河实验室 | In-vivo noninvasive brain function real-time monitoring system in hyperbaric oxygen chamber |
| CN118526161B (en)* | 2024-06-17 | 2024-11-15 | 脑机交互与人机共融海河实验室 | A non-invasive real-time monitoring system of brain function in a hyperbaric oxygen chamber |
| CN118963553A (en)* | 2024-07-30 | 2024-11-15 | 济宁医学院附属医院 | A patient physiological information monitoring system |
| Publication number | Publication date |
|---|---|
| CN108577830B (en) | 2021-10-26 |
| Publication | Publication Date | Title |
|---|---|---|
| CN108577830B (en) | User-oriented physical sign information dynamic monitoring method and system | |
| CN108478209B (en) | Electrocardio information dynamic monitoring method and dynamic monitoring system | |
| CN108309263A (en) | Multi-parameter monitoring data analysing method and multi-parameter monitoring system | |
| JP6567728B2 (en) | How to predict patient survival | |
| CN108309262A (en) | Multi-parameter monitoring data analysing method and multi-parameter monitor | |
| JP6721156B2 (en) | Biological information analysis device, system, and program | |
| US8706464B2 (en) | Health data dynamics, its sources and linkage with genetic/molecular tests | |
| TWI650737B (en) | Wearable device and method for evaluating possible occurrence of cardiac arrest | |
| Jin et al. | Predicting cardiovascular disease from real-time electrocardiographic monitoring: An adaptive machine learning approach on a cell phone | |
| US11992360B2 (en) | System and method for heart rhythm detection and reporting | |
| Hadjem et al. | An ECG monitoring system for prediction of cardiac anomalies using WBAN | |
| KR20190121237A (en) | Systematic apparatus for motion sensor and optical sensor based cardiac arrhythmia triage | |
| US20220167856A1 (en) | Lung function monitoring from heart signals | |
| Shah | Vital sign monitoring and data fusion for paediatric triage | |
| Lu et al. | Uncertainties in the analysis of heart rate variability: a systematic review | |
| Yu et al. | An unobtrusive stress recognition system for the smart office | |
| Charlton | Continuous respiratory rate monitoring to detect clinical deteriorations using wearable sensors | |
| Al-Kaysi et al. | Design of Ecg Monitoring System for Cardiovascular Patients Using Labview | |
| Kew et al. | Wearable patch-type ECG using ubiquitous wireless sensor network for healthcare monitoring application | |
| Bassiouni et al. | Combination of ECG and PPG signals for smart healthcare systems: Techniques, applications, and challenges | |
| Divyabharathi et al. | Analysis of Human Physiological Parameters Using Real-Time HRV Estimation from Acquired ECG Signals | |
| Braojos et al. | Level Accuracy of Automatic and Real Time Detection of Atrial Fibrillation with a New Wireless ECG Recorder (The SmartCardia) | |
| Nasim | Adaptation of the Segmented Beat Modulation Method to support diagnosis of cardiovascular disorders using electrocardiographic tracings acquired by wearable sensors | |
| Rajbhoj et al. | Human stress monitoring system using electrocardiogram | |
| Devaki et al. | Real-Time Heart Sound Analysis and Abnormality Detection using Deep Learning Algorithms in a Digital Stethoscope |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | Effective date of registration:20200511 Address after:16th Floor, Block A, 668 Shenzhuan Highway, Songjiang District, Shanghai, 20112 Applicant after:YOCALY INFORMATION SCIENCE & TECHNOLOGY Co.,Ltd. Address before:102200, Beijing Changping Changping District science and Technology Park, super Road, No. 3, building 37 Applicant before:Lepu Medical Technology (Beijing) Co.,Ltd. | |
| TA01 | Transfer of patent application right | ||
| CB02 | Change of applicant information | Address after:201612 16 / F, block a, no.668, Xinzhuan Road, Songjiang District, Shanghai Applicant after:Shanghai Lepu Yunzhi Technology Co.,Ltd. Address before:201612 16 / F, block a, no.668, Xinzhuan Road, Songjiang District, Shanghai Applicant before:YOCALY INFORMATION SCIENCE & TECHNOLOGY Co.,Ltd. | |
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |