技术领域Technical field
本发明涉及一种临床血糖管理质量控制系统，尤其涉及一种基于人工智能的方案偏离半定量评估的临床血糖管理质量控制系统，属于临床医疗质控领域。The invention relates to a clinical blood sugar management quality control system, in particular to a clinical blood sugar management quality control system based on artificial intelligence that deviates from semi-quantitative evaluation, and belongs to the field of clinical medical quality control.
背景技术Background technique
临床医疗的方案偏离是指临床医疗（包括相关的临床试验）由于某种原因未按或者未全部按规定的标准、要求进行临床医疗活动的现象。在各大医疗机构系统中，往往不能整合各类数据资源而进行整体分析，从而难以在日产的海量临床数据中识别出违规现象，从而在问题发生时才在海量数据中心寻找产生问题的蛛丝马迹。然而一些数据记录仍然停留在考勤、大型手术设备、化验设备的使用操作记录，病例记录这些宏观的数据上，虽然数据海量，但是仍然体现不出细节，尤其是对于住院的护理阶段的细节，比如给药时间、进餐规律、医疗用药是否合规，医师是否存在过度治疗，以及院外的患者是否按照规定进行给药以及作息时间观察。Deviation from the clinical medical treatment plan refers to the phenomenon that clinical medical treatment (including related clinical trials) fails to perform clinical medical activities in accordance with or not fully comply with the prescribed standards and requirements for some reason. In the systems of major medical institutions, it is often impossible to integrate various data resources for overall analysis, making it difficult to identify irregularities in the massive clinical data produced in Japan, and to only look for clues to the problem in the massive data center when the problem occurs. However, some data records still focus on macroscopic data such as attendance, operation records of major surgical equipment, laboratory equipment, and case records. Although the data is massive, it still cannot reflect the details, especially the details of the hospitalization stage of care, such as Dosing time, meal patterns, medical medication compliance, whether doctors are over-treating, and whether patients outside the hospital are taking medication as prescribed and observing their work and rest schedules.
临床血糖管理的方案偏离的监测其实质就是临床数据中对于违规的识别，是对于血糖监测数据与标准数据的对比结果判断。因此如何收集这些数据，如何组织这些数据，从而高效地、准确地做出判断结果，以及定位可疑数据是亟待解决的问题。The essence of monitoring deviations from clinical blood glucose management plans is the identification of violations in clinical data, which is the judgment of the comparison results between blood glucose monitoring data and standard data. Therefore, how to collect this data and how to organize this data to make judgment results efficiently and accurately and locate suspicious data are issues that need to be solved urgently.
此外，现有手术设备以及化验设备都大多数依靠进口，所依托的系统都各自为营，没有统一的数据记录系统，因此如何整合所有手术设备和化验设备数据记录系统也是当务之急。In addition, most of the existing surgical equipment and laboratory equipment are imported, and the systems they rely on are independent and there is no unified data recording system. Therefore, how to integrate the data recording system of all surgical equipment and laboratory equipment is also a top priority.
发明内容Contents of the invention
本发明为解决上述问题将考虑如下几个方面的解决方案：第一，考虑患者在院和院外的数据采集系统，第二，考虑血糖管理相关数据的分类图案化，第三，考虑基于图案化的人工智能识别以及问题数据的揪出。本发明的所有提及的系统是指硬件系统（如计算机）及可以由包括计算机的所述硬件系统上运行的非暂时性存储介质，其中存储了可以进行医疗数据和护理数据采集、记录、传输的程序。并且使得上述问题大部分得以满意解决。In order to solve the above problems, the present invention will consider the following solutions: first, consider the data collection system of patients in the hospital and outside the hospital; second, consider the classification and patterning of blood sugar management related data; third, consider the patterning based on Artificial intelligence identification and identification of problem data. All systems mentioned in the present invention refer to hardware systems (such as computers) and non-transitory storage media that can be run on the hardware systems including computers, which store medical data and nursing data that can be collected, recorded, and transmitted. program of. And most of the above problems can be solved satisfactorily.
鉴于上述考虑本发明提供一种基于人工智能的方案偏离半定量评估的临床血糖管理质量控制系统，包括计算机形式的在院系统、服务器、客户端，其中在院系统对患者在院的与血糖管理相关的医疗数据和/或护理数据进行数据收集并传输给服务器，所述服务器根据所收集的数据确定数据是否缺失，从而发出提醒信号给客户端，以提醒数据的及时记录，当所述数据收集完整之后所述服务器继续基于收集的完整的数据进行人工智能的识别，从而进一步识别方案偏离的问题数据。In view of the above considerations, the present invention provides a clinical blood sugar management quality control system based on artificial intelligence that deviates from semi-quantitative evaluation, including a computer-based in-hospital system, a server, and a client, wherein the in-hospital system monitors the blood sugar management of patients in the hospital. Relevant medical data and/or nursing data are collected and transmitted to the server. The server determines whether the data is missing based on the collected data, thereby sending a reminder signal to the client to remind the timely recording of the data. When the data is collected After completion, the server continues to perform artificial intelligence identification based on the complete data collected, thereby further identifying problem data that deviates from the plan.
其中，所述在院系统包括，考勤系统，手术设备操作记录系统，化验设备操作记录系统，医疗器械记录系统，用于记录医疗器械进出院以及在院配给以及使用的数据，药物记录系统，用于记录药物进出院以及在院配给以及使用的数据，以及医疗器械与药物使用情况记录系统。Among them, the in-hospital system includes an attendance system, a surgical equipment operation record system, a laboratory equipment operation record system, a medical device record system, which is used to record data on the entry and discharge of medical devices as well as in-hospital distribution and use, and a drug record system. It is used to record data on the entry and exit of drugs, as well as in-hospital dispensing and usage, as well as a recording system for medical equipment and drug usage.
所述考勤系统通过打卡和/或人脸识别记录出勤数据，所述手术设备操作记录系统通过手术设备上的操作历史以及受术者信息进行第一数据记录，所述化验设备操作记录系统通过化验设备上的试验数据以及受检者信息进行第二数据记录，医疗器械记录系统通过医疗器械包装检验通道进行识别进院数量，以及通过医疗器械出院通道进行出院数量的识别，所述检验通道和出院通道上通过传送带上方安装的图像采集装置进行进出院数量的识别（例如，通过预先训练的智能模型得到识别算法，在使用初期日常进出通道的同时顺便进行训练集的采集，并在使用一段时期期间不断训练，最后得到识别率高的算法模型。对于药物同理。），通过医疗器械或其包装上的电子标签扫描而记录一次在院配给以及使用的数据，所述药物记录系统通过药物包装检验通道进行识别进院数量，以及通过药物出院通道进行出院数量的识别，通过药物包装上的电子标签扫描而记录在院配给以及使用的数据，所述医疗器械与药物使用情况记录系统通过医护人员进行手动录入。The attendance system records attendance data through punch-in and/or face recognition, the surgical equipment operation recording system records the first data through the operation history on the surgical equipment and the patient information, and the laboratory equipment operation recording system records the first data through laboratory tests. The test data on the equipment and the subject information are recorded as second data. The medical device recording system identifies the number of admissions to the hospital through the medical device packaging inspection channel, and identifies the number of discharges through the medical device discharge channel. The inspection channel and discharge The image acquisition device installed above the conveyor belt is used to identify the number of entering and exiting the hospital on the channel (for example, the recognition algorithm is obtained through a pre-trained intelligent model, and the training set is collected while using the daily entry and exit channel in the initial stage, and during a period of use Continuous training, and finally an algorithm model with a high recognition rate is obtained. The same is true for drugs.), the data of in-hospital dispensing and use is recorded by scanning the electronic label on the medical device or its packaging, and the drug recording system passes the drug packaging inspection The channel identifies the number of admissions to the hospital, and the drug discharge channel identifies the number of discharges. The electronic labels on the drug packages are scanned to record the data on distribution and use in the hospital. The medical device and drug usage recording system is carried out by medical staff. Manual entry.
所述出勤数据、第一数据记录、第二数据记录、所有医疗器械和药物的进出院数量作为所述医疗数据，以及配给以及使用的数据、手动录入的数据作为所述护理数据都上传至服务器，当医疗器械与药物使用情况记录系统识别到手动录入缺失和电子标签扫描缺失即通过所述服务器向客户端发送手动录入缺失信号和电子标签扫描缺失信号，以提醒补录，并且根据用械和用药的医嘱时间和医嘱时间以外的规定时间内进行数据扫描以识别手动录入缺失和电子标签扫描缺失。The attendance data, the first data record, the second data record, the incoming and outgoing quantities of all medical equipment and drugs are uploaded to the server as the medical data, as well as the distribution and usage data, and manually entered data as the nursing data. , when the medical device and drug usage recording system recognizes that the manual entry is missing and the electronic label scanning is missing, it will send a manual entry missing signal and an electronic label scanning missing signal to the client through the server to remind the re-entry, and according to the equipment used and Data scanning is performed during the doctor's order time for medication and within the specified time outside of the doctor's order time to identify missing manual entries and missing electronic label scanning.
可以理解的是，通过上述数据的系统记录或者手动记录即能检测医疗活动中主要环节的医疗物资配给和使用状况，从而实现医疗数据的分类物理存储，通过手动记录仪规范用械和用药的规范，以确保方案偏离的实时监测，并通过电子标签扫描和手动记录缺失的记录频次对医护人员进行工作的考核，以督促其向方案规范化方向形成职业习惯。It is understandable that through the systematic recording or manual recording of the above data, the distribution and use of medical supplies in the main links of medical activities can be detected, thereby achieving the classified physical storage of medical data, and standardizing the use of equipment and medication through manual recorders. , to ensure real-time monitoring of deviations from the plan, and to assess the work of medical staff through electronic tag scanning and manual recording of missing record frequencies, so as to urge them to form professional habits in the direction of standardization of the plan.
所述客户端包括在院客户端以及院外客户端，所述在院客户端包括医师以及其他医护人员进行病历记录、医疗方案记录、医疗器械和药物配给记录的计算机系统，院外客户端包括医师以及其他医护人员以及病患或就诊者的移动通讯设备，所述服务器根据病患或就诊者的医疗器械和药物的使用医嘱要求而在规定的用械和/或用药的所述医嘱时间和医嘱时间以外的规定时间内向其他医护人员的移动通讯设备发送一次提醒短消息，并向所述病患或就诊者的移动通讯设备间歇性地发送提醒短消息，以使得医疗方和病患方之间进行器械和/或药物的使用提醒，所述病患或就诊者通过在移动通讯设备上进行确认而消除提醒短消息，并将消除的操作信息发送至服务器以让所述服务器知晓院外用械和/或用药的正常。The clients include in-hospital clients and out-of-hospital clients. The in-hospital clients include computer systems for doctors and other medical staff to record medical records, medical plan records, medical equipment and drug distribution records. The out-of-hospital clients include doctors and other medical staff. The mobile communication equipment of other medical staff and patients or patients. The server shall use the medical equipment and/or medicine according to the medical instructions of the patient or the patient at the specified time and time according to the medical instructions. Send reminder short messages once to the mobile communication devices of other medical staff outside the specified time, and send reminder short messages intermittently to the mobile communication devices of the patients or patients, so as to facilitate communication between the medical side and the patient. To remind the use of equipment and/or medicines, the patient or patient cancels the reminder short message by confirming on the mobile communication device, and sends the eliminated operation information to the server to let the server know about the equipment and/or equipment used outside the hospital. Or medication is normal.
优选地，所述医嘱时间以外的规定时间为0.5小时-6小时。Preferably, the specified time other than the doctor's prescribed time is 0.5 hours to 6 hours.
间歇性地发送提醒短消息的总时长为1-2min，间歇频率是每20-30秒提醒一次。The total duration of sending short reminder messages intermittently is 1-2 minutes, and the intermittent frequency is a reminder every 20-30 seconds.
优选地，所述移动通讯设备包括智能手机、笔记本电脑、平板电脑。Preferably, the mobile communication device includes a smart phone, a notebook computer, or a tablet computer.
因此，电子标签扫描和手动记录缺失信息也能上报至在院客户端，以便于医师进行查看以及必要时及时督促其他医护人员进行的相关记录以及促成器械和药物的配给和使用到位。从而实现医疗方、患者方、医疗物资方的三方的物资流、数据流的交流，从而实现了医疗的监督以及用于服务器进一步人工智能识别发现可能造成方案偏离的问题数据，从而半定量地给出质量控制方案。所谓半定量是指发现可能引起方案偏离的问题数据，而不是关注数据的试剂情况，包括数量，剂量，频次，参数指标等与具体药物的种类、数量、服用时间等是否科学配置的问题；另一方面，对于药物的医疗器械数量的异常也可以实现半定量的可能的药过度医疗情况，例如心血管支架的滥用，某一类药物的不且当的小病大医情况的定性辅助判断等。Therefore, missing information from electronic tag scanning and manual recording can also be reported to the hospital client, so that doctors can review it and, if necessary, promptly supervise the relevant records made by other medical staff and promote the distribution and use of equipment and drugs. This enables the exchange of material flows and data flows between the medical side, the patient side, and the medical supplies side, thereby realizing medical supervision and further artificial intelligence identification of the server to discover problem data that may cause deviations from the plan, thereby semi-quantitatively Develop a quality control plan. The so-called semi-quantitative refers to the discovery of problematic data that may cause deviations from the protocol, rather than focusing on the reagent conditions of the data, including quantity, dosage, frequency, parameter indicators, etc., as well as the type, quantity, taking time, etc. of specific drugs, whether they are scientifically configured; another On the one hand, the abnormal number of medical devices for drugs can also achieve semi-quantitative identification of possible medical over-medication situations, such as the abuse of cardiovascular stents, qualitative auxiliary judgment of inappropriate medical conditions for minor illnesses of a certain type of drug, etc. .
应当强调的是，间歇性地发送提醒短消息的总时长的设置不宜过长或者过短，过长则会由于院外病患厌烦而大概率都会进行消除操作而也都会进行服药，虽然能够起到大概率实现服药的作用，但一方面使得患者系统的不良体验，另一方面也不能试探出其是否存在的忘吃药习惯的情况。这是因为恰当的时长提醒会让患者注意到吃药，但是如果习惯性忘记吃药，也一定概率不会理会消除操作，从而服务器接收不到消除的操作信息，进而监测到忘吃药的可能性。而过短则（数秒到十几秒的区间内）一定概率会起不到提醒的效果。因此经过长期的观察确认1-2min是一个比较合适的提醒总时长。It should be emphasized that the total duration of sending reminder short messages intermittently should not be too long or too short. If it is too long, patients outside the hospital will most likely perform elimination operations and take medication due to boredom. Although it can play a role It has a high probability of achieving the effect of taking medicine, but on the one hand, it will cause the patient a negative experience in the system, and on the other hand, it cannot detect whether there is a habit of forgetting to take medicine. This is because an appropriate reminder will make the patient notice taking the medicine, but if he habitually forgets to take the medicine, there is a certain probability that the elimination operation will be ignored. As a result, the server will not receive the elimination operation information, and then the possibility of forgetting to take the medicine will be detected. sex. If it is too short (in the range of several seconds to more than ten seconds), there is a certain probability that the reminder will not be effective. Therefore, after long-term observation, it has been confirmed that 1-2 minutes is a more appropriate total reminder time.
所述人工智能的识别的方法包括如下步骤：The artificial intelligence identification method includes the following steps:
S1收集不同时期的所述医疗数据、以及每个患者和/或就诊者信息下的器械、药物、手术、化验的名称、器械和药物数量、服用数量和频次，以及使用数量和频次是否正常，其中所述使用数量和频次是否正常通过所述电子标签扫描、手动记录、消除的操作信息三项数据进行识别正常与否，缺少其中任一项均视作不正常；将所述医疗数据、每个患者和/或就诊者信息下的器械、药物、手术、化验的名称、数量、使用数量和频次，以及使用数量和频次是否正常的数据进行训练集、验证集的划分，两者比例为6-2:1；所述每个患者和/或就诊者信息包括患者和/或就诊者姓名、年龄、挂号科室、就医医师姓名；S1 collects the medical data of different periods, as well as the names of devices, drugs, surgeries, and tests under each patient and/or patient's information, the number of devices and drugs, the quantity and frequency of taking, and whether the quantity and frequency of use are normal, Whether the quantity and frequency of use are normal or not is identified through the three data of electronic tag scanning, manual recording, and deleted operation information. The absence of any one of these is regarded as abnormal; the medical data, each The name, quantity, usage quantity and frequency of equipment, medicines, surgeries and tests under individual patient and/or patient information, as well as the data on whether the usage quantity and frequency are normal are divided into training sets and validation sets. The ratio between the two is 6 -2:1; The information of each patient and/or patient includes the name, age, registration department, and name of the doctor;
S2按照患者和/或就诊者的就诊时间顺序进行排序，将其信息以及信息下的器械、药物、手术、化验的名称、数量、使用数量和频次，以及使用数量和频次是否正常的数据进行数据集合，并将医疗数据进行集合，形成数据集；将数据集中的数据按照规定进行彩色值的赋予的伪彩化处理，形成在医疗数据集对应的像素集之后按照时间顺序进行排序的像素集，形成集合图像；S2 sorts the patients and/or patients according to the order of their visit time, and collects data on their information and the name, quantity, usage quantity and frequency of the equipment, medicines, surgeries and tests under the information, as well as whether the quantity and frequency of usage are normal. Collect and assemble the medical data to form a data set; perform pseudo-colorization processing on the data in the data set by assigning color values according to regulations to form a pixel set that is sorted in chronological order after the pixel set corresponding to the medical data set. form a collective image;
S3根据所述集合图像，将质量控制系统预测的标准的数据对应的彩色值进行重新填充，形成标准集合图像，标准集合图像是指图像中的每个数据对应的像素值都是正常状态下的像素值；S3 refills the color values corresponding to the standard data predicted by the quality control system according to the set image to form a standard set image. The standard set image means that the pixel value corresponding to each data in the image is normal. Pixel values;
可以理解的是，标准的数据的确定应当由资历深的专家医师或者行业内的权威医师根据病患的病历和/或各项检测报告数据而进行确定，经过长期的病患的器械药物配给和使用方法确定之后，质量控制系统将会根据病历和/或各项检测报告数据学习到标准的数据，从而做出标准的数据的预测。It is understandable that the standard data should be determined by experienced expert physicians or authoritative physicians in the industry based on the patient's medical records and/or various test report data. After the usage method is determined, the quality control system will learn standard data based on medical records and/or various test report data, and then make standard data predictions.
S4将集合图像中对应所述医疗数据，以及每个患者和/或就诊者的信息下的器械、药物、手术、化验的名称、数量、使用数量和频次、使用数量和频次是否正常的数据对应的图像部分分别输入到医疗数据CNN模型以及患者和/或就诊者CNN模型中，两种CNN模型的输出端都通过softmax函数进行正常与否的二分分类输出，与实际的分类进行对比，从而根据验证集计算正确率以及计算损失函数，从而调整CNN网络参数，直到正确率以及损失函数稳定停止训练；S4 will collect data corresponding to the medical data in the collection image, as well as the name, quantity, usage quantity and frequency of equipment, drugs, surgeries, and tests under the information of each patient and/or patient, and whether the usage quantity and frequency are normal. The image part is input into the medical data CNN model and the patient and/or patient CNN model respectively. The output ends of both CNN models use the softmax function to perform dichotomous classification output of normal or not, and compare it with the actual classification, so as to The verification set calculates the accuracy and loss function, thereby adjusting the CNN network parameters until the accuracy and loss function are stable to stop training;
所述进一步识别方案偏离的问题数据进一步包括：The problem data for further identifying plan deviation further includes:
S5 将待预测的集合图像输入训练好的S4中的CNN模型中，当判断结果为不正常，则将所述待预测的集合图像与根据S3形成的对应的标准集合图像进行差分，得到差分结果就可以知道存在问题数据对应的患者和/或负责医师，最后服务器将判断和问题数据传送给在院系统。S5 Input the set image to be predicted into the trained CNN model in S4. When the judgment result is abnormal, the set image to be predicted is differentiated from the corresponding standard set image formed according to S3 to obtain the difference result. Then you can know the patient and/or responsible physician corresponding to the problematic data, and finally the server transmits the judgment and problem data to the hospital system.
优选地，所述CNN模型为基于残差机制的ResNET。Preferably, the CNN model is ResNET based on a residual mechanism.
应当理解的是，集合图像中所有像素点的大小一致，因数据的种类数据量不同而需要对于图像进行区域性划分，划分过程中可能存在不是均分的现象，比如后文实施例中出现的左右或者上下相邻的数据项目是多对一的，比如化验名称左侧有服用数量和服用频次两项数据项目，则像素点仍然会在这三项数据项目中大小一致地排布，此时对于化验名称而言将使用多个像素点来表示一个化验名称。It should be understood that the size of all pixels in the set image is the same. Due to the different types of data and the amount of data, the image needs to be divided into regions. There may be uneven divisions during the division process, such as what will appear in the embodiments below. The data items adjacent to the left and right or up and down are many-to-one. For example, if there are two data items on the left side of the test name, the number of doses and the frequency of doses, the pixels will still be arranged in the same size in these three data items. At this time For assay names, multiple pixels will be used to represent an assay name.
根据本发明，相比于现有技术显著存在至少如下三个方面的有益效果：According to the present invention, compared with the prior art, there are significant beneficial effects in at least the following three aspects:
1.通过各数据的记录系统实现了医疗物资的进出院以及配给使用记录。1. Through the recording system of each data, the entry and discharge of medical supplies as well as the distribution and use records are realized.
2.通过电子标签和手动记录实现了医疗物资使用正常化的督促和职业良好习惯培养，并通过医疗器械与药物使用情况记录系统以及服务器实现定点和移动式的双重提醒方式。2. Through electronic tags and manual records, we can supervise the normalization of the use of medical supplies and cultivate good professional habits, and realize fixed-point and mobile dual reminders through the medical equipment and drug usage recording system and servers.
3.通过人工智能预测以及标准集合图像的差分识别出存在问题数据对应的患者和/或负责医师，实现了宏观与微观的数据的图案化质量控制方案，将海量的数据组织到一张集合图像中，方便了数据的组织与处理。3. Through artificial intelligence prediction and the difference of standard collection images, the patient and/or responsible physician corresponding to the problematic data is identified, a patterned quality control scheme for macro and micro data is implemented, and massive data are organized into a collection image. , which facilitates the organization and processing of data.
附图说明Description of drawings
图1本发明实施例1中一种基于人工智能的方案偏离半定量评估的临床血糖管理质量控制系统结构示意图。Figure 1 is a schematic structural diagram of a clinical blood glucose management quality control system in which the artificial intelligence-based solution deviates from semi-quantitative assessment in Embodiment 1 of the present invention.
图2本发明实施例2中集合图像的内部构成示意图。Figure 2 is a schematic diagram of the internal structure of the collective image in Embodiment 2 of the present invention.
图3本发明实施例2中的集合图像中图像部分1和图像部分2内部结构以及CNN模型训练过程流程示意图。Figure 3 is a schematic flowchart of the internal structure of image part 1 and image part 2 in the set image and the CNN model training process in Embodiment 2 of the present invention.
图4问题数据发现的差分图像方法的示意图。Figure 4 Schematic diagram of the differential image method for problem data discovery.
其中附图标记，1不同时期的医疗数据对应的图像部分，2患者/就诊者对应的图像部分。The reference numerals are: 1. image parts corresponding to medical data in different periods; 2. image parts corresponding to patients/visitors.
具体实施方式Detailed ways
下面将参照附图更详细地描述本发明的示例性实施例。虽然附图中显示了本发明的示例性实施例，然而应当理解，可以以各种形式实现本发明而不应被这里阐述的实施例所限制。相反，提供这些实施例是为了能够更透彻地理解本发明，并且能够将本发明的范围完整的传达给本领域的技术人员。Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a thorough understanding of the invention, and to fully convey the scope of the invention to those skilled in the art.
实施例1Example 1
本实施例将说明一种基于人工智能的方案偏离半定量评估的临床血糖管理质量控制系统，如图1所示，一种基于人工智能的方案偏离半定量评估的临床血糖管理质量控制系统，包括在院系统、服务器、客户端，其中在院系统对在院的与血糖管理相关的医疗数据和护理数据进行数据收集从而传输给服务器，所述服务器根据所收集的数据而确定数据是否缺失，从而发出提醒信号给客户端，以提醒数据的及时记录，当所述数据收集完整之后所述服务器继续基于收集的完整的数据进行人工智能的识别，从而进一步识别方案偏离的问题数据，将判断结果和问题数据发送给在院系统，以在在院系统中的相应的记录系统中进行显示提醒。This embodiment will illustrate a clinical blood glucose management quality control system based on artificial intelligence for semi-quantitative assessment of protocol deviations. As shown in Figure 1, a clinical blood glucose management quality control system based on artificial intelligence for semi-quantitative assessment of protocol deviations includes In-hospital system, server, and client, where the in-hospital system collects data on medical data and nursing data related to blood sugar management in the hospital and transmits it to the server. The server determines whether the data is missing based on the collected data, thereby Send a reminder signal to the client to remind the timely recording of data. After the data collection is complete, the server continues to perform artificial intelligence identification based on the complete data collected, thereby further identifying problem data that deviates from the plan, and combines the judgment results with The problem data is sent to the hospital system for display and reminder in the corresponding recording system in the hospital system.
其中，所述在院系统的记录系统包括了考勤系统，手术设备操作记录系统，化验设备操作记录系统，医疗器械记录系统，用于记录医疗器械进出院以及在院配给以及使用的数据，药物记录系统，用于记录药物进出院以及在院配给以及使用的数据，以及医疗器械与药物使用情况记录系统。Among them, the recording system of the hospital system includes an attendance system, a surgical equipment operation recording system, a laboratory equipment operation recording system, and a medical equipment recording system, which are used to record data on the entry and discharge of medical equipment as well as in-hospital distribution and use, and drug records. The system is used to record the data on the entry and discharge of drugs, as well as the distribution and use in the hospital, as well as the medical device and drug usage recording system.
所述考勤系统通过人脸识别记录出勤数据，所述手术设备操作记录系统通过手术设备上的操作历史以及受术者信息进行第一数据记录，所述化验设备操作记录系统通过化验设备上的试验数据以及受检者信息进行第二数据记录，医疗器械记录系统通过医疗器械包装检验通道进行识别进院数量，以及通过医疗器械出院通道进行出院数量的识别，所述检验通道和出院通道上通过传送带上方安装的图像采集装置进行进出院数量的识别，通过医疗器械或其包装上的电子标签扫描而记录一次在院配给以及使用的数据，所述药物记录系统通过药物包装检验通道进行识别进院数量，以及通过药物出院通道进行出院数量的识别，通过药物包装上的电子标签扫描而记录在院配给以及使用的数据，所述医疗器械与药物使用情况记录系统通过医护人员进行手动录入。The attendance system records attendance data through face recognition, the surgical equipment operation recording system records the first data through the operation history on the surgical equipment and the patient information, and the laboratory equipment operation recording system records tests on the laboratory equipment. The data and subject information are recorded in the second data. The medical device recording system identifies the number of admissions to the hospital through the medical device packaging inspection channel, and identifies the number of discharges through the medical device discharge channel. The inspection channel and discharge channel are passed through a conveyor belt. The image acquisition device installed above identifies the number of admissions and discharges, and records the data of distribution and use in the hospital by scanning the electronic label on the medical device or its packaging. The drug recording system identifies the number of admissions to the hospital through the drug packaging inspection channel. , and identify the discharge quantity through the drug discharge channel, and record the distribution and use data in the hospital by scanning the electronic label on the drug package. The medical device and drug usage recording system is manually entered by medical staff.
所述出勤数据、第一数据记录、第二数据记录、所有医疗器械和药物的进出院数量作为所述医疗数据，以及配给以及使用的数据、手动录入的数据作为所述护理数据都上传至服务器，当医疗器械与药物使用情况记录系统识别到手动录入缺失和电子标签扫描缺失即通过所述服务器向客户端发送手动录入缺失信号和电子标签扫描缺失信号，以提醒补录，并且根据用械和用药的医嘱时间和医嘱时间以外的规定时间内进行数据扫描以识别手动录入缺失和电子标签扫描缺失。The attendance data, the first data record, the second data record, the incoming and outgoing quantities of all medical equipment and drugs are uploaded to the server as the medical data, as well as the distribution and usage data, and manually entered data as the nursing data. , when the medical device and drug usage recording system recognizes that the manual entry is missing and the electronic label scanning is missing, it will send a manual entry missing signal and an electronic label scanning missing signal to the client through the server to remind the re-entry, and according to the equipment used and Data scanning is performed during the doctor's order time for medication and within the specified time outside of the doctor's order time to identify missing manual entries and missing electronic label scanning.
所述客户端包括在院客户端以及院外客户端，所述在院客户端包括医师以及其他医护人员进行病历记录、医疗方案记录、医疗器械和药物配给记录的计算机系统，院外客户端包括医师以及其他医护人员以及病患或就诊者的移动通讯设备，所述服务器根据病患或就诊者的医疗器械和药物的使用医嘱要求而在规定的用械和/或用药的所述医嘱时间和医嘱时间以外的规定时间内向其他医护人员的移动通讯设备发送一次提醒短消息，并向所述病患或就诊者的移动通讯设备间歇性地发送提醒短消息，以使得医疗方和病患方之间进行器械和/或药物的使用提醒，所述病患或就诊者通过在移动通讯设备上进行确认而消除提醒短消息，并将消除的操作信息发送至服务器以让所述服务器知晓院外用械和/或用药的正常。The clients include in-hospital clients and out-of-hospital clients. The in-hospital clients include computer systems for doctors and other medical staff to record medical records, medical plan records, medical equipment and drug distribution records. The out-of-hospital clients include doctors and other medical staff. The mobile communication equipment of other medical staff and patients or patients. The server shall use the medical equipment and/or medicine according to the medical instructions of the patient or the patient at the specified time and time according to the medical instructions. Send reminder short messages once to the mobile communication devices of other medical staff outside the specified time, and send reminder short messages intermittently to the mobile communication devices of the patients or patients, so as to facilitate communication between the medical side and the patient. To remind the use of equipment and/or medicines, the patient or patient cancels the reminder short message by confirming on the mobile communication device, and sends the eliminated operation information to the server to let the server know about the equipment and/or equipment used outside the hospital. Or medication is normal.
所述医嘱时间以外的规定时间为0.5小时-6小时，在另一个较佳实施例中，所述医嘱时间以外的规定时间为1小时。间歇性地发送提醒短消息的总时长为1-2min，间歇频率是每20-30秒提醒一次。在另一个较佳实施例中，间歇性地发送提醒短消息的总时长为1min，间歇频率是每25秒提醒一次。所述移动通讯设备包括智能手机、笔记本电脑、平板电脑。在另一个较佳实施例中，所述移动通讯设备包括智能手机。 The prescribed time other than the doctor's prescribed time is 0.5 hours to 6 hours. In another preferred embodiment, the prescribed time other than the doctor's prescribed time is 1 hour. The total duration of sending short reminder messages intermittently is 1-2 minutes, and the intermittent frequency is a reminder every 20-30 seconds. In another preferred embodiment, the total duration of sending reminder short messages intermittently is 1 minute, and the intermittent frequency is a reminder every 25 seconds. The mobile communication devices include smart phones, laptops, and tablet computers. In another preferred embodiment, the mobile communication device includes a smartphone.
实施例2Example 2
本实施例将说明采用上述实施例1的临床血糖管理质量控制系统进行人工智能的识别的方法以及所述进一步识别方案偏离的问题数据的步骤：其中说明所述人工智能的识别的方法包括：This embodiment will illustrate the method of artificial intelligence identification using the clinical blood glucose management quality control system of the above-mentioned embodiment 1 and the steps of further identifying problem data that deviates from the plan: the method of identifying the artificial intelligence includes:
S1收集不同时期的所述医疗数据（与患者血糖管理相关的医疗数据）、以及每个患者和/或就诊者信息下的器械、药物、手术、化验的名称、器械和药物数量、服用数量和频次，以及使用数量和频次是否正常，其中所述使用数量和频次是否正常通过所述电子标签扫描、手动记录、消除的操作信息三项数据进行识别正常与否，缺少其中任一项均视作不正常；将所述医疗数据、每个患者和/或就诊者信息下的器械、药物、手术、化验的名称、数量、使用数量和频次，以及使用数量和频次是否正常的数据进行训练集、验证集的划分，两者比例为3:1；所述每个患者和/或就诊者信息包括患者和/或就诊者姓名、年龄、挂号科室、就医医师姓名。S1 collects the medical data (medical data related to patient blood sugar management) at different periods, as well as the names of devices, drugs, surgeries, tests, the number of devices and drugs, the number of doses taken, and Frequency, and whether the quantity and frequency of use are normal. Whether the quantity and frequency of use are normal or not can be identified through the three data of electronic tag scanning, manual recording, and elimination of operation information. The absence of any one of these is regarded as Abnormal; create a training set based on the medical data, the name, quantity, usage quantity and frequency of equipment, medicines, surgeries and tests under each patient and/or patient information, as well as whether the usage quantity and frequency are normal, For the division of the verification set, the ratio between the two is 3:1; the information of each patient and/or patient includes the name, age, registration department, and name of the doctor.
S2如图2所示，按照患者和/或就诊者的就诊时间顺序即箭头方向按序进行排序，将其信息以及信息下的器械、药物、手术、化验的名称、数量、使用数量和频次，以及使用数量和频次是否正常的数据进行数据集合形成图像部分2（包含多个按箭头时间顺序排序的图像部分），并将医疗数据进行集合形成图像部分1，形成数据集；将数据集中的数据按照规定进行彩色值的赋予的伪彩化处理，形成图像部分1之后按照时间顺序进行排序的像素集，形成集合图像；图3显示了图像部分1和图像部分2的内部构造。As shown in Figure 2, S2 sorts the patients and/or patients in the order of their visit time, that is, in the direction of the arrow, and lists their information and the name, quantity, usage quantity and frequency of the instruments, drugs, surgeries, and tests under the information. And use data with normal quantity and frequency to form image part 2 (containing multiple image parts sorted in arrow time order), and collect medical data to form image part 1 to form a data set; combine the data in the data set After the pseudo-colorization process of assigning color values according to regulations, image part 1 is formed, and then a set of pixels is sorted in time order to form a collective image; Figure 3 shows the internal structure of image part 1 and image part 2.
S3根据所述集合图像，将质量控制系统预测的标准的数据对应的彩色值进行重新填充，形成标准集合图像。S3 refills the color values corresponding to the standard data predicted by the quality control system according to the set image to form a standard set image.
S4如图3所示，将集合图像中对应所述医疗数据，以及每个患者和/或就诊者的信息下的器械、药物、手术、化验的名称、数量、使用数量和频次、使用数量和频次是否正常的数据对应的图像部分1和图像部分2分别输入到医疗数据CNN模型以及患者和/或就诊者CNN模型中，两种CNN模型的输出端都通过softmax函数进行正常与否的二分分类输出，与实际的分类进行对比，从而根据验证集计算正确率以及计算损失函数，从而调整CNN网络参数，直到正确率以及损失函数稳定停止训练。S4, as shown in Figure 3, collects the medical data corresponding to the medical data in the collection image, as well as the name, quantity, usage quantity and frequency, use quantity and Image part 1 and image part 2 corresponding to the data of whether the frequency is normal are input into the medical data CNN model and the patient and/or patient CNN model respectively. The output ends of both CNN models use the softmax function to perform dichotomous classification of normal or not. The output is compared with the actual classification to calculate the accuracy and loss function based on the verification set, thereby adjusting the CNN network parameters until the accuracy and loss function are stable to stop training.
所述进一步识别方案偏离的问题数据的步骤包括：The steps of further identifying problematic data that deviate from the plan include:
S5 将待预测的集合图像输入训练好的S4中的CNN模型中，当判断结果为不正常，如图4所示，则将所述待预测的集合图像与根据S3形成的对应的标准集合图像进行差分，得到差分结果就可以知道存在问题数据对应的患者和/或负责医师，最后服务器将判断和问题数据传送给在院系统。S5 Input the set image to be predicted into the trained CNN model in S4. When the judgment result is abnormal, as shown in Figure 4, the set image to be predicted is compared with the corresponding standard set image formed according to S3. Perform a difference and get the difference result to know the patient and/or responsible physician corresponding to the problematic data. Finally, the server transmits the judgment and problem data to the hospital system.
这里图3中的所述CNN模型为基于残差机制的ResNET。The CNN model in Figure 3 here is ResNET based on the residual mechanism.
最后应说明的是：以上各实施例仅用以说明本发明的技术方案，而非对其限制；尽管参照前述各实施例对本发明进行了详细的说明，本领域的普通技术人员应当理解：在本发明的精神和原则之内，其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分或者全部技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案脱离本发明的保护范围。Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: Within the spirit and principles of the present invention, it is still possible to modify the technical solutions recorded in the foregoing embodiments, or to make equivalent substitutions for some or all of the technical features; and these modifications or substitutions do not deviate from the corresponding technical solutions. protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577574.0ACN117292807B (en) | 2023-11-24 | 2023-11-24 | A clinical blood glucose management quality control system |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577574.0ACN117292807B (en) | 2023-11-24 | 2023-11-24 | A clinical blood glucose management quality control system |
| Publication Number | Publication Date |
|---|---|
| CN117292807A CN117292807A (en) | 2023-12-26 |
| CN117292807Btrue CN117292807B (en) | 2024-02-06 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311577574.0AActiveCN117292807B (en) | 2023-11-24 | 2023-11-24 | A clinical blood glucose management quality control system |
| Country | Link |
|---|---|
| CN (1) | CN117292807B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1999058050A1 (en)* | 1998-05-13 | 1999-11-18 | Cygnus, Inc. | Signal processing for measurement of physiological analytes |
| CN105361892A (en)* | 2015-12-17 | 2016-03-02 | 无锡桑尼安科技有限公司 | Personnel physiological state recognition platform based on race detection |
| WO2021207016A1 (en)* | 2020-04-05 | 2021-10-14 | Theator inc. | Systems and methods for automating video data management during surgical procedures using artificial intelligence |
| CN115312205A (en)* | 2022-10-11 | 2022-11-08 | 四川省医学科学院·四川省人民医院 | Single-disease clinical path management system |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230072368A1 (en)* | 2019-10-03 | 2023-03-09 | Rom Technologies, Inc. | System and method for using an artificial intelligence engine to optimize a treatment plan |
| US20210118559A1 (en)* | 2019-10-22 | 2021-04-22 | Tempus Labs, Inc. | Artificial intelligence assisted precision medicine enhancements to standardized laboratory diagnostic testing |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO1999058050A1 (en)* | 1998-05-13 | 1999-11-18 | Cygnus, Inc. | Signal processing for measurement of physiological analytes |
| CN105361892A (en)* | 2015-12-17 | 2016-03-02 | 无锡桑尼安科技有限公司 | Personnel physiological state recognition platform based on race detection |
| WO2021207016A1 (en)* | 2020-04-05 | 2021-10-14 | Theator inc. | Systems and methods for automating video data management during surgical procedures using artificial intelligence |
| CN115312205A (en)* | 2022-10-11 | 2022-11-08 | 四川省医学科学院·四川省人民医院 | Single-disease clinical path management system |
| Publication number | Publication date |
|---|---|
| CN117292807A (en) | 2023-12-26 |
| Publication | Publication Date | Title |
|---|---|---|
| World Health Organization | Classification of digital interventions, services and applications in health: a shared language to describe the uses of digital technology for health | |
| US20200051679A1 (en) | Methods and systems for a pharmacological tracking and reporting platform | |
| US20160378950A1 (en) | Method and apparatus for tracking a pharmaceutical | |
| CN109599188A (en) | A kind of medical care information management system | |
| CN108932968A (en) | A kind of clinical trial subjects management system | |
| US20140324216A1 (en) | Method, system and apparatus for managing medication delivery | |
| CN103718208A (en) | Method and device for monitoring medication compliance | |
| CN117012364A (en) | Medical health service cloud platform based on industrial Internet technology | |
| WO2002002004A9 (en) | A health outcomes and disease management network and related method for providing improved patient care | |
| US20110015943A1 (en) | Comprehensive method and system for intake screening and medical records management | |
| US11631484B1 (en) | Method and apparatus for predicting, encouraging, and intervening to improve patient medication adherence | |
| US20200152305A1 (en) | Healthcare compliance process over a network | |
| CN110111910A (en) | The management of radiotherapy patient intelligent follow-up and QA system | |
| US11289179B1 (en) | Automated medication compliance assurance system | |
| CN105447655A (en) | Clinical medical care quality management system | |
| CN111640475A (en) | Management system for clinical test | |
| CN118866283A (en) | A method and system for full-course diagnosis and treatment based on artificial intelligence | |
| US20160292386A1 (en) | Systems and architecture for electronic interfaces and complex data structures for medication reconciliation and patient regimen adherence detection | |
| CN116130119A (en) | Breast cancer postoperative rehabilitation auxiliary management system | |
| Karakolias | Mapping data-driven strategies in improving health care and patient satisfaction | |
| CN112863698A (en) | Chronic disease follow-up tracking method | |
| CN118748094A (en) | A method for reviewing adverse drug reactions | |
| Wei et al. | [Retracted] Clinical Care of Hyperthyroidism Using Wearable Medical Devices in a Medical IoT Scenario | |
| Gorelova et al. | A Discrete Event Simulation of patient flow in an Assisted Reproduction Clinic with the integration of a smart health monitoring system | |
| Fung et al. | Comparison of electronic pharmacy prescription records with manually collected medication histories in an emergency department |
| 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |