WO2021151329A1

Movatterモバイル変換

Info

Publication number: WO2021151329A1
Application number: PCT/CN2020/124261
Authority: WO
Inventors: 左磊; 赵惟; 徐卓扬; 廖希洋; 赵婷婷; 孙行智
Original assignee: Ping An Technology Shenzhen Co Ltd
Current assignee: Ping An Technology Shenzhen Co Ltd
Priority date: 2020-09-09
Filing date: 2020-10-28
Publication date: 2021-08-05
Anticipated expiration: 2023-03-09
Also published as: CN112037932B; CN112037932A

Abstract

A patient medication behavior intervention method and apparatus, and a server and a storage medium, which are applicable to medical technology. The method comprises: acquiring behavior data of a target patient within a preset duration, wherein the behavior data comprises non-medication behavior data, and the preset duration comprises n time periods (S101); learning the behavior data by means of a long short term memory (LSTM) network so as to obtain a medication behavior probability sequence of an (n+1)th time period (S102); determining, according to the medication behavior probability sequence, a medication behavior type of the target patient within the (n+1)th time period (S103); and determining a target intervention strategy according to the medication behavior type and the non-medication behavior data, and sending medication reminding information to the target patient on the basis of the target intervention strategy (S104). By means of the method, the medication compliance of a patient can be improved.

Description

Translated fromChinese

患者用药行为干预方法及装置、服务器、存储介质Patient medication behavior intervention method and device, server and storage medium

本申请要求于2020年9月9日提交中国专利局、申请号为202010940492.8，发明名称为“患者用药行为干预方法及装置、服务器、存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。This application claims the priority of a Chinese patent application filed with the Chinese Patent Office on September 9, 2020, the application number is 202010940492.8, and the invention title is "Patient medication behavior intervention methods and devices, servers, and storage media", the entire content of which is incorporated by reference Incorporated in this application.

技术领域Technical field

本申请涉及医疗卫生服务互联网技术领域，尤其涉及一种患者用药行为干预方法及装置、服务器、存储介质。This application relates to the field of medical and health services Internet technology, and in particular to a method and device for intervention in patient medication behavior, a server, and a storage medium.

背景技术Background technique

用药依从性较差是慢病管理中常见的问题，导致这种问题产生的原因主要如下：患者对用药依从的认识不到位，认为病情好转即可停药，如高血压患者一旦发现血压恢复正常则停药；临床医生对患者的患教力度不够，患者没有掌握相关知识，存在忘记服药、剂量错误等情况。Poor medication compliance is a common problem in chronic disease management. The main reasons for this problem are as follows: patients do not have a good understanding of medication compliance, and they can stop medication if they think their condition improves. For example, once hypertensive patients find that their blood pressure returns to normal The drug is stopped; the clinician is not sufficiently educating the patient on the patient, the patient does not have relevant knowledge, and there are situations such as forgetting to take the drug or wrong dosage.

发明人意识到，针对患者的用药干预通常是在患者停药或不按时服药后身体产生不适，再次就医时才进行干预，干预不及时。此外，为了管理大量患者，现有技术的用药干预通常只是简单针对药物分类进行干预，干预效果不佳，适用性差。The inventor realizes that medication intervention for patients usually involves discomfort after the patient stops taking the medication or does not take the medication on time, and the intervention is only performed when the patient goes to the doctor again, and the intervention is not timely. In addition, in order to manage a large number of patients, medication interventions in the prior art usually simply intervene based on drug classification, and the intervention effect is not good and the applicability is poor.

发明内容Summary of the invention

本申请提供一种患者用药行为干预方法及装置、服务器、存储介质，可提高患者用药依从性，适用性高。This application provides a method and device, a server, and a storage medium for intervention in patient medication behavior, which can improve patient medication compliance and have high applicability.

第一方面，本申请提供了一种患者用药行为干预方法，包括：In the first aspect, this application provides a method for intervention of patients' medication behavior, including:

获取目标患者在预设时长内的行为数据，上述行为数据包括非用药行为数据，上述预设时长中包括n个时间段，n为自然数；Obtain behavior data of the target patient within a preset time period, where the behavior data includes non-drug behavior data, and the preset time period includes n time periods, where n is a natural number;

通过长短时记忆网络LSTM对上述行为数据进行学习，以获得第n+1时间段的用药行为概率序列；Learning the above-mentioned behavior data through the long and short-term memory network LSTM to obtain the probability sequence of medication behavior in the n+1th time period;

根据上述用药行为概率序列确定上述目标患者在上述第n+1时间段的用药行为类型；Determine the medication behavior type of the target patient in the n+1th time period according to the probability sequence of medication behavior;

根据上述用药行为类型和上述非用药行为数据确定目标干预策略，并基于上述目标干预策略向上述目标患者发送用药提醒信息。Determine the target intervention strategy according to the above-mentioned medication behavior type and the above-mentioned non-medication behavior data, and send medication reminder information to the above-mentioned target patient based on the above-mentioned target intervention strategy.

第二方面，本申请提供了一种患者用药行为干预装置，包括：In the second aspect, this application provides a patient medication behavior intervention device, including:

行为数据获取模块，用于获取目标患者在预设时长内的行为数据，上述行为数据包括非用药行为数据，上述预设时长中包括n个时间段，n为自然数；The behavioral data acquisition module is used to acquire behavioral data of the target patient within a preset time period, the above-mentioned behavioral data includes non-medicine behavior data, the above-mentioned preset time period includes n time periods, and n is a natural number;

概率序列学习模块，用于通过长短时记忆网络LSTM对上述行为数据进行学习，以获得第n+1时间段的用药行为概率序列；The probability sequence learning module is used to learn the above-mentioned behavior data through the long and short-term memory network LSTM to obtain the probability sequence of medication behavior in the n+1th time period;

行为类型确定模块，用于根据上述用药行为概率序列确定上述目标患者在上述第n+1时间段的用药行为类型；The behavior type determination module is used to determine the medication behavior type of the target patient in the n+1th time period according to the probability sequence of medication behavior;

确定发送模块，用于根据上述用药行为类型和上述非用药行为数据确定目标干预策略，并基于上述目标干预策略向上述目标患者发送用药提醒信息。The determining and sending module is used to determine a target intervention strategy according to the above-mentioned medication behavior type and the above-mentioned non-medication behavior data, and to send medication reminder information to the above-mentioned target patient based on the above-mentioned target intervention strategy.

第三方面，本申请提供了一种服务器，包括处理器、存储器和收发器，上述处理器、存储器和收发器相互连接，其中，上述存储器用于存储计算机程序，上述计算机程序包括程序指令；上述处理器被配置用于调用上述程序指令，执行以下方法：In a third aspect, the present application provides a server including a processor, a memory, and a transceiver. The processor, the memory, and the transceiver are connected to each other, wherein the memory is used to store a computer program, and the computer program includes program instructions; The processor is configured to call the above program instructions and execute the following methods:

根据上述用药行为类型和上述非用药行为数据确定目标干预策略，并基于上述目标干预策略向上述目标患者发送用药提醒信息。Determine the target intervention strategy based on the above-mentioned medication behavior types and the above-mentioned non-medication behavior data, and send medication reminders to the above-mentioned target patients based on the above-mentioned target intervention strategy.

第四方面，本申请提供了一种计算机可读存储介质，上述计算机可读存储介质存储有计算机程序，上述计算机程序包括程序指令；上述程序指令当被处理器执行时使上述处理器执行以下方法：In a fourth aspect, this application provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and the computer program includes program instructions; when the program instructions are executed by a processor, the processor executes the following method :

采用本申请，可以提高患者用药依从性。With this application, the compliance of patients with medication can be improved.

附图说明Description of the drawings

为了更清楚地说明本申请的技术方案，下面将对实施例描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to explain the technical solution of the present application more clearly, the following will briefly introduce the drawings used in the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. Ordinary technicians can obtain other drawings based on these drawings without creative work.

图1是本申请提供的用药行为干预系统的架构示意图；Figure 1 is a schematic diagram of the architecture of the medication behavior intervention system provided by this application;

图2是本申请提供的患者用药行为干预方法的一流程示意图；Fig. 2 is a schematic flow chart of a method for intervention of patient medication behavior provided by the present application;

图3是本申请提供的患者用药行为干预方法的另一流程示意图；Figure 3 is a schematic diagram of another flow chart of the intervention method for patient medication behavior provided by the present application;

图4是本申请提供的患者用药行为干预装置的结构示意图；Fig. 4 is a schematic structural diagram of a patient medication behavior intervention device provided by the present application;

图5是本申请提供的服务器的结构示意图。Fig. 5 is a schematic diagram of the structure of the server provided by the present application.

具体实施方式Detailed ways

下面将结合本申请中的附图，对本申请中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative work shall fall within the protection scope of this application.

本申请的技术方案可应用于人工智能、数字医疗、智慧城市、区块链和/或大数据技术领域，以提升患者用药依从性，实现智慧医疗。The technical solution of this application can be applied to the fields of artificial intelligence, digital medicine, smart city, blockchain and/or big data technology to improve patient medication compliance and realize smart medical treatment.

本申请提出一种患者用药行为干预方法，可根据目标患者在前n个时间段的行为数据，通过长短时记忆(long short-term memory，LSTM)网络模型对前n个时间段的行为数据进行学习确定目标患者第n+1时间段的用药行为类型，根据用药行为类型和前n个时间段的行为数据中的非用药行为数据确定目标干预策略，并基于目标干预策略向目标患者发送用药提醒信息，提高了患者用药依从性。This application proposes a medication behavior intervention method for patients, which can perform the behavior data of the first n time periods through a long short-term memory (LSTM) network model based on the behavior data of the target patient in the first n time periods. Learn to determine the medication behavior type of the target patient in the n+1 time period, determine the target intervention strategy based on the medication behavior type and the non-drug behavior data in the behavior data of the previous n time periods, and send medication reminders to the target patient based on the target intervention strategy Information improves the compliance of patients with medication.

本申请提供的患者用药行为干预方法可适用于用药行为干预系统，该系统中包括用药行为干预平台和患者终端集群，请参见图1，是本申请提供的用药行为干预系统的架构示意图。如图1所示，该构架示意图包括用药行为干预平台100和患者终端集群101，其中，患者终端集群101中可以包括多个患者终端，如图1所示，具体可以包括患者终端101a、患者终端101b、患者终端101c、…、患者终端101n，目标患者终端可以为患者终端集群101中的任意一个患者终端，本申请以图1中的终端患者101a为目标患者终端进行说明。The patient medication behavior intervention method provided in this application can be applied to a medication behavior intervention system, which includes a medication behavior intervention platform and a patient terminal cluster. Please refer to FIG. 1, which is a schematic diagram of the architecture of the medication behavior intervention system provided in this application. As shown in FIG. 1, the schematic diagram of the architecture includes a medication behavior intervention platform 100 and a patient terminal cluster 101. The patient terminal cluster 101 may include multiple patient terminals. As shown in FIG. 1, it may specifically include a patient terminal 101a and a patient terminal. The patient terminal 101b, the patient terminal 101c, ..., the patient terminal 101n, the target patient terminal may be any patient terminal in the patient terminal cluster 101, and this application takes the terminal patient 101a in FIG. 1 as the target patient terminal for description.

其中，用药行为干预平台100和患者终端集群101中每个患者终端可以为计算机设备，包括手机、平板电脑、笔记本电脑、掌上电脑、移动互联网设备(MID，mobile internet device)、销售点(Point Of Sales，POS)机、可穿戴设备(例如智能手表、智能手环等)等。Among them, each patient terminal in the medication behavior intervention platform 100 and the patient terminal cluster 101 can be a computer device, including a mobile phone, a tablet computer, a notebook computer, a handheld computer, a mobile internet device (MID, mobile internet device), and a point of sale (Point Of Sale). Sales, POS) machines, wearable devices (such as smart watches, smart bracelets, etc.), etc.

在本申请提供的患者用药行为干预方法中，用药行为干预平台100可按照预设频率向患者终端101a发送行为数据采集提示以提示患者终端101a的目标患者按照目标数据采集形式反馈行为数据。这里，预设频率可为一天一次、一周一次等，目标数据采集形式可包括网页链接采集或二维码采集，本申请对此不做限制。用药行为干预平台100接收患者终端101a反馈的行为数据作为目标患者在前n个时间段的行为数据。这里，n为自然数，行为数据可包括用药行为数据和非用药行为数据，其中，用药行为数据可包括药品数据、服药数据和停药数据等，非用药行为数据可包括患者基本数据、生活方式数据和身体症状数据等。用药行为干预平台100可通过LSTM对目标患者在前n个时间段的行为数据进行学习，从而获得第n+1时间段的用药行为概率序列，并将用药行为概率序列中用药行为概率最大值对应的用药行为类型确定为目标患者在第n+1时间段的用药行为类型。之后，用药行为干预平台100根据用药行为类型和非用药行为数据生成自我管理标签，计算自我管理标签与预设干预策略集合中各个干预策略的策略标签之间的匹配度值，根据匹配度值确定目标干预策略，并根据目标干预策略向目标患者所在的患者终端101a发送用药提醒信息。In the patient medication behavior intervention method provided in this application, the medication behavior intervention platform 100 can send behavior data collection prompts to the patient terminal 101a at a preset frequency to prompt the target patient of the patient terminal 101a to feed back behavior data in the form of target data collection. Here, the preset frequency can be once a day, once a week, etc. The target data collection form can include web link collection or QR code collection, which is not limited in this application. The medication behavior intervention platform 100 receives the behavior data fed back by the patient terminal 101a as the behavior data of the target patient in the first n time periods. Here, n is a natural number, and behavioral data can include medication behavior data and non-drug behavior data. Among them, medication behavior data can include drug data, medication data, and drug withdrawal data, etc., and non-drug behavior data can include patient basic data and lifestyle data. And physical symptoms data, etc. The medication behavior intervention platform 100 can learn the behavior data of the target patient in the first n time periods through LSTM, so as to obtain the medication behavior probability sequence of the n+1th time period, and correspond to the maximum value of the medication behavior probability in the medication behavior probability sequence The medication behavior type of is determined as the medication behavior type of the target patient in the n+1 time period. After that, the medication behavior intervention platform 100 generates a self-management label according to the medication behavior type and non-medicine behavior data, calculates the matching degree value between the self-management label and the strategy label of each intervention strategy in the preset intervention strategy set, and determines the matching degree value Target intervention strategy, and send medication reminder information to the patient terminal 101a where the target patient is located according to the target intervention strategy.

为方便描述，下面将以用药行为干预平台为执行主体，结合图2至图3对本申请提供的患者用药行为干预方法进行示例说明。For the convenience of description, the following will take the medication behavior intervention platform as the main body of execution, combined with Figures 2 to 3 to illustrate the patient medication behavior intervention methods provided in this application.

参见图2，是本申请提供的患者用药行为干预方法的一流程示意图。如图2所示，本申请提供的方法可包括如下步骤：Refer to FIG. 2, which is a schematic flow chart of the method for intervention of patient medication behavior provided by the present application. As shown in Figure 2, the method provided by this application may include the following steps:

S101，获取目标患者在预设时长内的行为数据。S101: Obtain behavior data of a target patient within a preset time period.

在一些可行的实施方式中，上述预设时长中包括n个时间段，n为自然数。比如预设时长可以是一个月，一个时间段可以是一周。上述行为数据可包括用药行为数据和非用药行为数据。其中，用药行为数据可包括药品数据(如苯磺酸氨氯地平片，5mg/次，1次/天)、服药数据(如忘记吃药，忘记吃药1次)和停药数据(如有意停药)等。非用药行为数据可包括患者基本数据(如年龄、性别、种族、身高、体重)、生活方式数据(如是否抽烟、是否饮酒等)和身体症状数据(如血压水平变化、有无头晕、有无视线模糊等)。In some feasible implementation manners, the foregoing preset duration includes n time periods, and n is a natural number. For example, the preset duration can be one month, and one time period can be one week. The above-mentioned behavior data may include drug use behavior data and non-drug use behavior data. Among them, the medication behavior data can include medication data (such as amlodipine besylate tablets, 5 mg/time, once/day), medication data (such as forgetting to take medicine, forgetting to take medicine once) and drug withdrawal data (if intentionally Withdrawal) and so on. Non-drug behavior data can include basic patient data (such as age, gender, race, height, weight), lifestyle data (such as whether to smoke, whether to drink alcohol, etc.), and physical symptoms data (such as changes in blood pressure levels, whether dizziness, whether or not Blurred vision, etc.).

在一些可行的实施方式中，用药行为干预平台按照预设频率向目标患者发送行为数据采集提示以提示目标患者按照目标数据采集形式反馈行为数据，接收目标患者反馈的行为数据以作为目标患者在预设时长内的行为数据。其中，目标数据采集形式包括网页链接采集或二维码采集。In some feasible implementations, the medication behavior intervention platform sends behavior data collection prompts to the target patient at a preset frequency to prompt the target patient to feed back behavior data in the form of target data collection, and receives the behavior data fed back by the target patient to serve as the target patient in the forecast. Set the behavior data within the time period. Among them, the target data collection form includes web link collection or QR code collection.

举例来说，假设预设时长为七月至八月共两个月，即八周，预设频率为一周一次，用药行为干预平台可在七月至八月中每周五下午七点向高血压患者A发送行为数据采集提示。高血压患者A接收该行为数据采集提示，并通过访问行为数据采集提示携带的网页链接去填写每周的行为数据，并在填写完成后通过点击网页链接所在网页的确认按键，向用药行为干预平台返回每周的行为数据。用药行为干预平台接收高血压患者A七月至八月每周的行为数据，从而得到高血压患者A在七月至八月中八周的行为数据。For example, suppose that the default duration is two months from July to August, that is, eight weeks, and the default frequency is once a week. The medication behavior intervention platform can be higher at 7 pm every Friday from July to August. Blood pressure patient A sends a behavioral data collection reminder. Hypertensive patient A receives the behavioral data collection reminder, and fills in weekly behavioral data by visiting the webpage link carried by the behavioral data collection reminder, and after completing the filling, clicks the confirmation button on the webpage where the webpage link is located to report to the medication behavior intervention platform Returns weekly behavior data. The medication behavior intervention platform receives the weekly behavior data of hypertensive patient A from July to August, and obtains the behavior data of hypertensive patient A during the eight weeks from July to August.

S102，通过长短时记忆网络LSTM对行为数据进行学习，以获得第n+1时间段的用药行为概率序列。S102, learning the behavior data through the long and short-term memory network LSTM, to obtain the medication behavior probability sequence in the n+1th time period.

在一些可行的实施方式中，用药行为干预平台可首先构建一个LSTM网络模型，将预设时长内的行为数据按照预设顺序进行排序以得到排序后的行为数据序列，并将行为数据序列输入LSTM。通过LSTM以多种用药行为类型的多分类问题为学习任务来学习n个时间段中各个时间段的行为数据对应的下一时间段的用药行为类型，从而获得第n+1时间段的用药行为概率序列。其中，预设顺序可为时间从早到晚的顺序，用药行为类型可为有意停药、忘记吃药1次、忘记吃药2-3次和忘记吃药3次以上四种类型，具体可根据实际应用场景确定，在此不做限定。时间段的形式可为一天、一周等，具体可根据实际应用场景确定，在此不做限定。In some feasible implementations, the medication behavior intervention platform can first construct an LSTM network model, sort the behavior data within a preset time period in a preset order to obtain the sorted behavior data sequence, and input the behavior data sequence into the LSTM . Through the LSTM, the multi-classification problem of multiple medication behavior types is used as the learning task to learn the medication behavior type of the next time period corresponding to the behavior data of each time period in n time periods, so as to obtain the medication behavior of the n+1th time period Probability sequence. Among them, the preset order can be from morning to night, and the medication behavior types can be intentionally stopping medication, forgetting to take medication once, forgetting to take medication 2-3 times, and forgetting to take medication more than 3 times. It is determined according to the actual application scenario and is not limited here. The time period can be in the form of one day, one week, etc., which can be specifically determined according to actual application scenarios, and is not limited here.

举例来说，用药行为干预平台将7月份4周的行为数据(生活行为数据、用药行为类型等)按照第一周、第二周、第三周和第四周的时间顺序进行排序后得到行为数据序列。之后，将行为数据序列输入LSTM，通过LSTM以有意停药、忘记吃药1次、忘记吃药2-3次和忘记吃药3次以上四种用药行为类型的多分类问题为学习任务，来学习4周中第一周的行为数据对应的第二周的用药行为类型、…、第三周的行为数据对应的第四周的用药行为类型，并得到第五周的用药行为概率序列。For example, the medication behavior intervention platform sorts the behavior data (life behavior data, medication behavior type, etc.) for the 4 weeks of July in the order of the first week, second week, third week, and fourth week to get the behavior Data sequence. After that, input the behavior data sequence into LSTM. Through LSTM, the multi-classification problem of four types of medication behaviors: intentionally stopping the drug, forgetting to take the drug once, forgetting to take the drug 2-3 times, and forgetting to take the drug 3 times or more are the learning tasks. Learn the medication behavior type of the second week corresponding to the behavior data of the first week of 4 weeks,..., the medication behavior type of the fourth week corresponding to the behavior data of the third week, and obtain the medication behavior probability sequence of the fifth week.

在一些可行的实现方式中，本申请实施例提供的双向LSTM中包括多个LSTM记忆单元，其中，LSTM记忆单元中各参数可通过如下公式1至5确定。In some feasible implementation manners, the bidirectional LSTM provided in the embodiments of the present application includes a plurality of LSTM memory units, where each parameter in the LSTM memory unit can be determined by the following formulas 1 to 5.

其中，公式1至5满足：Among them, formulas 1 to 5 satisfy:

i_t＝σ(W_ix_t+U_ih_t-1) (1)i_t =σ(W_i x_t +U_i h_t-1 ) (1)

f_t＝σ(W_fx_t+U_fh_t-1) (2)f_t =σ(W_f x_t +U_f h_t-1 ) (2)

o_t＝σ(W_ox_t+U_oh_t-1) (3)o_t =σ(W_o x_t +U_o h_t-1 ) (3)

在上述公式1至5中，σ(x)与

均为非线性激活函数。In the above formulas 1 to 5, σ(x) and

All are non-linear activation functions.

其中，σ(x)为sigmoid函数并且满足：σ(x)＝(1+exp(-x))^-1。Among them, σ(x) is a sigmoid function and satisfies: σ(x)=(1+exp(-x))^-1 .

为tanh函数并且满足：

Is the tanh function and satisfies:

在本申请中，n个时间段的行为数据按照预设时间顺序(从早到晚)，通过排序的方式串联成为一个行为数据序列输入到LSTM中，因此，在某一时刻t输入的行为数据则对应于n个时间段中的某个时间段，因此，在上述公式1至5中，变量t可对应于时间段。x_t则表示对应于时刻t上输入的对应于时间段的行为数据。i_t，f_t和o_t分别代表时刻t的输入门，记忆门和输出门输出在时刻t输入的行为数据所对应的时间段的下一个时间段目标患者的用药行为概率序列。举例来说，若x_t表示输入的目标患者在第一周的行为数据，则o_t表示目标患者在第二周的用药行为概率序列。其中，上述输入门、记忆门和输出门统称为LSTM记忆单元的逻辑门。c_t表示时刻t输入的行为数据所表示的目标患者的信息，为方便描述可称为LSTM记忆单元在当前时刻t的信息。In this application, the behavior data of n time periods are in a preset time sequence (from morning to night), and they are serialized into a behavior data sequence and input into the LSTM in a sorting manner. Therefore, the behavior data input at a certain time t It corresponds to a certain time period among n time periods. Therefore, in the above formulas 1 to 5, the variable t may correspond to a time period. x_t represents the behavior data corresponding to the time period input at time t. i_t , f_t and o_t respectively represent the input gate at time t, and the memory gate and output gate output the drug behavior probability sequence of the target patient in the next time period corresponding to the behavior data input at time t. For example, if x_t represents the input behavior data of the target patient in the first week, then o_t represents the probability sequence of the medication behavior of the target patient in the second week. Among them, the above-mentioned input gate, memory gate and output gate are collectively referred to as the logic gates of the LSTM memory cell. c_t represents the information of the target patient represented by the behavior data input at time t, which can be referred to as the information of the LSTM memory unit at the current time t for the convenience of description.

在本申请提供的LSTM网络中，对于LSTM记忆单元在当前时刻t的信息以及LSTM记忆单元中每一个逻辑门(输入门、输出门、记忆门)所输出的概率的计算中均分别存在一个在当前时刻t对应于各个时间段的输入x_t和上一时刻t-1对应于各个时间段的上一时间段的隐含变量h_t-1的权重转移矩阵W。例如，对应i_t的W_i，对应于f_t的W_f，对应于o_t的W_o以及对应于c_t的W_c等。其中，上述隐含变量h_t-1可由上一时刻t-1输出门和记忆单元的输出确定。其中，隐含变量是隐形状态变量，是相对于可观测变量的参量。可观测变量可包括可以直接从待检测图像中得到的特征，隐含变量是高于这些可观测变量一层的抽象概念的变量，并隐含变量是可以用于控制可观测变量的变化的参量。In the LSTM network provided by this application, there is one of the information of the LSTM memory unit at the current time t and the calculation of the probability of each logic gate (input gate, output gate, memory gate) in the LSTM memory unit. The current time t corresponds to the input x_{t of} each time period and the previous time t-1 corresponds to the weight transfer matrix W_{of the implicit variable h t-1 of the previous time period of each time period.} For example, the corresponding_i W i_t, which corresponds to the f_t W_f, corresponding to the W_o o_t c_t and the corresponding W_c and the like. Among them, the above-mentioned hidden variable h_t-1 can be determined by the output of the output gate and the memory unit at the previous time t-1. Among them, the hidden variable is an invisible state variable, which is a parameter relative to the observable variable. Observable variables can include features that can be directly obtained from the image to be detected. Hidden variables are variables with an abstract concept higher than these observable variables, and hidden variables are parameters that can be used to control changes in observable variables. .

S103，根据用药行为概率序列确定目标患者在第n+1时间段的用药行为类型。S103: Determine the medication behavior type of the target patient in the n+1th time period according to the medication behavior probability sequence.

在一些可行的实施方式中，用药行为概率序列中一个用药行为概率对应一个用药行为，用药行为干预平台将用药行为概率序列中的最大用药行为概率所对应的用药行为确定为第n+1时间段的用药行为类型。In some feasible implementations, one medication behavior probability in the medication behavior probability sequence corresponds to one medication behavior, and the medication behavior intervention platform determines the medication behavior corresponding to the maximum medication behavior probability in the medication behavior probability sequence as the n+1th time period. Type of medication behavior.

举例来说，根据步骤S102得到第五周的用药行为概率序列为0.1，0.2，0.1，0.6，并且，该用药行为概率序列中的用药行为概率对应的用药行为分别为有意停药，忘记吃药1次，忘记吃药2-3次和忘记吃药3次以上，则用药行为干预平台将0.6对应的忘记吃药3次以上确定为第五周的用药行为类型。For example, according to step S102, the probability sequence of medication behavior in the fifth week is 0.1, 0.2, 0.1, 0.6, and the medication behavior corresponding to the probabilities of medication behavior in the probability sequence of medication behavior are intentional withdrawal and forgetting to take medication. Once, forget to take medicine 2-3 times and forget to take medicine more than 3 times, the medication behavior intervention platform will determine the medication behavior type for the fifth week of forgetting to take medication for more than 3 times corresponding to 0.6.

S104，根据用药行为类型和非用药行为数据确定目标干预策略，并基于目标干预策略向目标患者发送用药提醒信息。S104: Determine a target intervention strategy according to the medication behavior type and non-medicine behavior data, and send medication reminder information to the target patient based on the target intervention strategy.

在执行步骤S104之前，用药行为干预平台获取大量的用药行为干预方案，每个干预方案均包括患者的用药行为类型、非用药行为数据，以及干预策略，根据每个干预方案中患者的用药行为类型和非用药行为数据生成多个策略标签，或一个组合策略标签。例如，第一干预方案中高血压患者的用药行为类型为忘记吃药1次，非用药行为数据包括有运动行为和血压水平不变化，则上述方式生成的组合策略标签为“忘记吃药1次-运动-血压水平不变化”。之后，用药行为干预平台根据每个干预方案中的干预策略以及该干预策略携带的策略标签得到干预策略集合。Before performing step S104, the medication behavior intervention platform obtains a large number of medication behavior intervention programs. Each intervention program includes the patient's medication behavior type, non-medicine behavior data, and intervention strategy, according to the patient's medication behavior type in each intervention program And non-medicine behavior data to generate multiple strategy tags, or a combination strategy tag. For example, in the first intervention plan, the medication behavior type of hypertensive patients is forgetting to take medication once, and the non-medication behavior data includes exercise behavior and blood pressure level unchanged, then the combined strategy generated by the above method is labeled as "forgetting to take medication once- Exercise-the blood pressure level does not change". After that, the medication behavior intervention platform obtains the intervention strategy set according to the intervention strategy in each intervention plan and the strategy label carried by the intervention strategy.

需要说明的是，用药行为干预平台生成的策略标签和自我管理标签的形式相同，换句话说，若策略标签的形式为组合标签，则自我管理标签的形式也为组合标签；若策略标签的形式为多个标签，则自我管理标签的形式也为多个标签。本申请以策略标签和自我管理标签的形式均为多个标签进行举例说明。It should be noted that the form of the strategy label generated by the medication behavior intervention platform and the self-management label are the same. In other words, if the form of the strategy label is a combination label, the form of the self-management label is also a combination label; if the form of the strategy label is If there are multiple labels, the self-management label is also in the form of multiple labels. In this application, both the policy label and the self-management label are multiple labels for illustration.

在一些可行的实施方式中，用药行为干预平台根据用药行为类型和非用药行为数据生成目标患者的自我管理标签，计算自我管理标签与干预策略集合中各个干预策略的策略标签的匹配度以得到多个匹配度值，预设干预策略集合包括多个干预策略，一个干预策略携带至少一个策略标签；将多个匹配度值中最大值对应的干预策略确定为目标干预策略。In some feasible implementations, the medication behavior intervention platform generates the self-management label of the target patient based on the medication behavior type and non-drug behavior data, and calculates the degree of matching between the self-management label and the strategy label of each intervention strategy in the intervention strategy set to obtain multiple The preset intervention strategy set includes multiple intervention strategies, and one intervention strategy carries at least one strategy tag; the intervention strategy corresponding to the maximum value among the multiple matching degree values is determined as the target intervention strategy.

具体的，用药行为干预平台可根据用药行为类型，非用药行为数据中的生活方式数据和身体症状数据生成自我管理标签。例如，若目标患者的用药行为类型为有意停药，生活方式数据包括无抽烟行为和有饮酒行为，身体症状数据为血压水平变化范围大，则根据上述数据生成的自我管理标签可为有意停药、饮酒、血压水平变化范围大。若目标患者的自我管理标签为A、B、C，第i个干预策略的策略标签为a、b、c，则自我管理标签与第i个干预策略的策略标签之间的匹配度计算公式可以为自我管理标签A、B、C分别与第i个干预策略的策略标签a、b、c之间的匹配度的总和。其中，单个自我管理标签与单个策略标签之间的匹配度可以通过预设的匹配度表格(如表1所示，表1为自我管理标签与策略标签的匹配表)获得。Specifically, the medication behavior intervention platform can generate self-management labels based on the type of medication behavior, lifestyle data and physical symptom data in non-medicine behavior data. For example, if the target patient’s medication behavior type is intentional withdrawal, lifestyle data includes non-smoking behaviors and drinking behaviors, and the physical symptom data indicates that the blood pressure level varies widely, then the self-management label generated based on the above data can be intentional withdrawal , Drinking, and blood pressure levels vary widely. If the target patient's self-management labels are A, B, C, and the strategy labels of the i-th intervention strategy are a, b, c, the calculation formula for the matching degree between the self-management label and the strategy label of the i-th intervention strategy can be It is the sum of the matching degrees between the self-management labels A, B, and C respectively and the strategy labels a, b, and c of the i-th intervention strategy. Among them, the matching degree between a single self-management label and a single policy label can be obtained through a preset matching degree table (as shown in Table 1, which is a matching table between the self-management label and the policy label).

表1Table 1

其中，自我管理标签中的A、B、C、D、E、……可分别表示有意停药、忘记吃药3次以上、忘记吃药2-3次、忘记吃药1次、血压变化范围大、……，策略标签中的a、b、c、……可分别表示有意停药、忘记吃药3次以上、忘记吃药2-3次、……，并且，策略标签a与自我管理标签A、B、C、D、E之间的匹配度分别为10、0、0、0、5，策略标签b与自我管理标签A、B、C、D、E之间的匹配度分别为10、0、0、0、4，策略标签c与自我管理标签A、B、C、D、E之间的匹配度分别为10、0、0、0、3。Among them, A, B, C, D, E, ... in the self-management label can respectively indicate intentional withdrawal, forgetting to take medicine more than 3 times, forgetting to take medicine 2-3 times, forgetting to take medicine once, and the range of blood pressure change. Big,..., a, b, c, ... in the strategy label can respectively indicate intention to stop the drug, forget to take the drug more than 3 times, forget to take the drug 2-3 times, ..., and the strategy label a and self-management The matching degrees between tags A, B, C, D, and E are 10, 0, 0, 0, and 5, respectively. The matching degrees between policy tag b and self-management tags A, B, C, D, and E are respectively 10, 0, 0, 0, 4, the matching degree between the policy label c and the self-management label A, B, C, D, E is 10, 0, 0, 0, 3, respectively.

例如，假设高血压患者X的自我管理标签包括忘记吃药2-3次、饮酒、血压水平变化范围较大，第i个干预策略的策略标签包括有意停药、抽烟、血压水平变化范围大。用药行为干预平台分别通过查找预设的匹配度表格(如表1所示)的方式获得自我管理标签“忘记吃药2-3次”分别与策略标签“有意停药”、“抽烟”和“血压水平变化范围大”之间的匹配度为0、0和3，自我管理标签“饮酒”分别与策略标签“有意停药”、“抽烟”和“血压水平变化范围大”之间的匹配度为0、0和2，自我管理标签“血压水平变化范围大”分别与策略标签“有意停药”、“抽烟”和“血压水平变化范围大”之间的匹配度为5、1和10，则自我管理标签与第i个干预策略的策略标签之间的匹配度为21。根据上述方式计算得到自我管理标签与干预策略集合中各个干预策略的策略标签之间的匹配度，从而得到多个匹配度值，将多个匹配度值中最大值对应的干预策略确定为目标干预策略。For example, suppose that the self-management label of hypertensive patient X includes forgetting to take medicine 2-3 times, drinking alcohol, and a large range of blood pressure level changes. The strategy label of the i-th intervention strategy includes intentional drug withdrawal, smoking, and a large range of blood pressure level changes. The medication behavior intervention platform obtains the self-management labels “forgot to take medicine 2-3 times” by searching the preset matching degree table (shown in Table 1), respectively, and the strategy labels “intentionally stop medication”, “smoking” and “ The matching degree between “Big blood pressure level change range” is 0, 0, and 3, and the matching degree between the self-management label “drinking” and the strategy label “Intentionally stopping drug”, “Smoking” and “Big blood pressure level change range” respectively Is 0, 0, and 2, and the matching degrees between the self-management label "Big range of blood pressure level change" and the strategy labels "Intentionally stopping medication", "Smoking" and "Big range of blood pressure level change" are 5, 1, and 10. Then the matching degree between the self-management label and the strategy label of the i-th intervention strategy is 21. According to the above method, the matching degree between the self-management label and the strategy label of each intervention strategy in the intervention strategy set is calculated to obtain multiple matching degree values, and the intervention strategy corresponding to the maximum value among the multiple matching degree values is determined as the target intervention Strategy.

在一些可行的实施方式中，目标干预策略包括用药提醒方式和用药提醒频率。用药行为干预平台在预设时间段内，按照用药提醒频率以用药提醒方式向目标患者发送用药提醒信息，其中，用药提醒方式包括文字提示或语音提示。可选的，文字提示可包括短信提示、邮件提示、APP提示等，语音提示可包括电话提示。In some feasible implementation manners, the target intervention strategy includes medication reminder mode and medication reminder frequency. The medication behavior intervention platform sends medication reminder information to the target patient in a medication reminder manner according to the medication reminder frequency within a preset time period, where the medication reminder method includes text prompts or voice prompts. Optionally, the text prompts may include short message prompts, email prompts, APP prompts, etc., and the voice prompts may include phone prompts.

举例来说，用药行为干预平台分别在每天的早上九点以及晚上七点，以短信形式向高血压患者A发送包含提醒用药以及高血压并发症方面内容的短信。For example, the medication behavior intervention platform sends text messages to hypertensive patient A in the form of text messages at 9 am and 7 pm each day, including reminders of medication and complications of hypertension.

在本申请中，用药行为干预平台可根据目标患者在n个时间段的行为数据，通过LSTM得到目标患者在第n+1时间段的用药行为概率序列，进而确定目标患者在第n+1时间段的用药行为，并根据目标患者在n个时间段的行为数据中的非用药行为数据和第n+1时间段的用药行为确定目标干预策略，基于目标干预策略向目标患者发送用药提醒信息，从而使干预更加及时，提高了患者用药依从性。In this application, the medication behavior intervention platform can obtain the medication behavior probability sequence of the target patient in the n+1 time period through LSTM based on the behavior data of the target patient in n time periods, and then determine the target patient in the n+1 time period To determine the target intervention strategy based on the non-drug behavior data in the behavior data of the target patient in n time periods and the drug behavior in the n+1 time period, and send medication reminders to the target patient based on the target intervention strategy. This makes the intervention more timely and improves the patient's medication compliance.

请参见图3，是本申请提供的患者用药行为干预方法的另一流程示意图。如图3所示，本申请提供的方法可包括如下步骤：Please refer to Fig. 3, which is a schematic diagram of another process of the intervention method for patient medication behavior provided in this application. As shown in Figure 3, the method provided by this application may include the following steps:

S201，获取目标患者在预设时长内的行为数据，行为数据包括非用药行为数据，预设时长中包括n个时间段。S201: Obtain behavior data of the target patient within a preset time period, where the behavior data includes non-medicine behavior data, and the preset time period includes n time periods.

S202，通过长短时记忆网络LSTM对行为数据进行学习，以获得第n+1时间段的用药行为概率序列。S202: Learning the behavior data through the long- and short-term memory network LSTM to obtain the medication behavior probability sequence in the n+1th time period.

S203，根据用药行为概率序列确定目标患者在第n+1时间段的用药行为类型。S203: Determine the medication behavior type of the target patient in the n+1th time period according to the medication behavior probability sequence.

在一些可行的实施方式中，上述步骤S201至步骤S203所执行的实现方式可参见上述图2所示实施例中步骤S101至S103所提供的实现方式，在此不再赘述。In some feasible implementation manners, the implementation manners performed by the foregoing steps S201 to S203 can refer to the implementation manners provided by steps S101 to S103 in the embodiment shown in FIG. 2, and details are not described herein again.

S204，根据用药行为类型和非用药行为数据确定目标干预策略，目标干预策略包括用药提醒方式和用药提醒频率。S204: Determine a target intervention strategy according to the medication behavior type and non-medicine behavior data. The target intervention strategy includes a medication reminder mode and a medication reminder frequency.

在一些可行的实施方式中，在执行步骤S204之前，用药行为干预平台可根据临床医学知识和某种疾病患者的行为数据构建某种疾病的知识图谱，该知识图谱包括某种疾病患者的多个干预策略，每个干预策略包括患者的至少一个组合标签，其中，组合标签可由实体-属性形式表示。实体可包括生活方式、病情衡量指标、用药行为类型等，生活方式实体的属性可包括抽烟、饮酒、运动等。这里，以高血压患者对病情衡量指标进行举例，病情衡量指标可为血压水平变化，血压水平变化的属性包括收缩压下限值、收缩压上限值、舒张压下限值和舒张压上限值，用药行为类型的属性可包括有意停药、忘记吃药1次、忘记吃药2-3次和忘记吃药3次以上。这里，实体的属性可以以分数的形式表示，例如，用药行为类型的属性：有意停药、忘记吃药1次、忘记吃药2-3次和忘记吃药3次以上，可分别用0、3、2、1的分数表示。之后，用药行为干预平台根据目标患者的非用药行为数据和用药行为类型，在知识图谱中遍历每个干预策略的组合标签，直至找到与目标患者非用药行为数据和用药行为类型匹配度最高的干预策略，并将该干预策略确定为目标干预策略。In some feasible implementations, before step S204 is performed, the medication behavior intervention platform can construct a knowledge map of a certain disease based on clinical medical knowledge and behavior data of patients with a certain disease, and the knowledge map includes multiple knowledge maps of patients with a certain disease. Intervention strategies, each intervention strategy includes at least one combination label of the patient, where the combination label can be represented in an entity-attribute form. Entities can include lifestyle, disease measurement indicators, medication behavior types, etc., and attributes of lifestyle entities can include smoking, drinking, exercise, and so on. Here, taking a hypertensive patient as an example of the disease measurement index, the disease measurement index can be the change of blood pressure level. The attributes of the blood pressure level change include the lower limit of systolic blood pressure, the upper limit of systolic blood pressure, the lower limit of diastolic blood pressure, and the upper limit of diastolic blood pressure. Value, the attributes of the medication behavior type can include intentionally stopping the medication, forgetting to take the medication once, forgetting to take the medication 2-3 times, and forgetting to take the medication more than 3 times. Here, the attributes of the entity can be expressed in the form of scores. For example, the attributes of the medication behavior type: intentionally stop the medication, forget to take the medication once, forget to take the medication 2-3 times, and forget to take the medication more than 3 times, which can be 0, The scores of 3, 2, and 1 are expressed. After that, the medication behavior intervention platform traverses the combined label of each intervention strategy in the knowledge graph according to the target patient's non-medication behavior data and medication behavior type, until it finds the intervention with the highest matching degree with the target patient's non-medication behavior data and medication behavior type Strategy and determine the intervention strategy as the target intervention strategy.

S205，在预设时间段内，按照用药提醒频率以用药提醒方式向目标患者发送用药提醒信息。S205, within a preset time period, send medication reminder information to the target patient in a medication reminder manner according to the medication reminder frequency.

其中，预设时间段和用药提醒频率可由患者所服用药物的时间和服用频率确定，此外，用药提醒信息除了包含提醒患者用药的消息外，还包括针对目标患者的非用药行为数据和用药行为类型所提供的患教内容。Among them, the preset time period and medication reminder frequency can be determined by the time and frequency of medication taken by the patient. In addition, medication reminder information includes not only a message reminding the patient to take medication, but also non-medication behavior data and medication behavior type for the target patient The content of patient education provided.

举例来说，若目标患者所服用药物的时间为饭前，服用频率为一日三次，非用药行为数据包括抽烟，并且在下一周的用药行为类型为故意停药，则用药行为干预平台分别在每天的早上八点、中午十一点和晚上六点以电话的形式提醒目标患者服用药物，并对目标患者进行用药依从性的患教。For example, if the target patient takes the drug before meals, the frequency of the drug is three times a day, the non-medication behavior data includes smoking, and the medication behavior type in the next week is deliberate withdrawal, then the medication behavior intervention platform will be used every day At 8 o’clock in the morning, 11 o’clock noon and 6 o’clock in the evening, the target patients were reminded to take medication by telephone, and the target patients were educated on medication compliance.

在本申请中，用药行为干预平台可根据目标患者在n个时间段的行为数据，通过LSTM得到目标患者在第n+1时间段的用药行为概率序列，进而确定目标患者在第n+1时间段的用药行为，并根据目标患者在n个时间段的行为数据中的非用药行为数据和第n+1时间段的用药行为确定目标干预策略。该目标干预策略包括用药提醒频率和用药提醒方式，在预设时间段内，按照用药提醒频率以用药提醒方式向目标患者发送用药提醒信息，从而使干预更加及时，提高了患者用药依从性。此外，目标干预策略均是根据患者的行为数据确定的，目标干预策略中的用药提醒信息包括患教内容，从而不仅实现了对患者用药行为的个性化干预，还减轻了医生的压力，提高了慢病管理质量。In this application, the medication behavior intervention platform can obtain the medication behavior probability sequence of the target patient in the n+1 time period through LSTM based on the behavior data of the target patient in n time periods, and then determine the target patient in the n+1 time period To determine the target intervention strategy based on the non-drug behavior data in the behavior data of the target patient in the n time period and the medication behavior in the n+1 time period. The target intervention strategy includes medication reminder frequency and medication reminder mode. Within a preset time period, medication reminder information is sent to the target patient in the medication reminder mode according to the medication reminder frequency, so that the intervention is more timely and the patient's medication compliance is improved. In addition, the target intervention strategy is determined based on the patient’s behavioral data. The medication reminder information in the target intervention strategy includes the content of patient education, which not only realizes the personalized intervention of the patient’s medication behavior, but also reduces the pressure on the doctor and improves the slowness. Quality of disease management.

基于上述方法实施例的描述，本申请还提供了一种患者用药行为干预装置，该患者用药行为干预装置可以是上述方法实施例中的用药行为干预平台。请参见图4，是本申请提供的一种患者用药行为干预装置的结构示意图。如图4所示，该患者用药行为干预装置4可以包括：行为数据获取模块41、概率序列学习模块42、行为类型确定模块43和确定发送模块44。Based on the description of the foregoing method embodiment, the present application also provides a patient medication behavior intervention device. The patient medication behavior intervention device may be the medication behavior intervention platform in the foregoing method embodiment. Please refer to FIG. 4, which is a schematic structural diagram of a patient medication behavior intervention device provided by the present application. As shown in FIG. 4, the patient medicationbehavior intervention device 4 may include: a behaviordata acquisition module 41, a probability sequence learning module 42, a behaviortype determination module 43, and adetermination sending module 44.

行为数据获取模块41，用于获取目标患者在预设时长内的行为数据，上述行为数据包括非用药行为数据，上述预设时长中包括n个时间段，n为自然数；The behaviordata acquisition module 41 is configured to acquire behavior data of the target patient within a preset time period, where the above-mentioned behavior data includes non-drug use behavior data, and the above-mentioned preset time period includes n time periods, and n is a natural number;

概率序列学习模块42，用于通过长短时记忆网络LSTM对上述行为数据进行学习，以获得第n+1时间段的用药行为概率序列；The probability sequence learning module 42 is used to learn the above-mentioned behavior data through the long and short-term memory network LSTM to obtain the probability sequence of medication behavior in the n+1th time period;

行为类型确定模块43，用于根据上述用药行为概率序列确定上述目标患者在上述第n+1时间段的用药行为类型；The behaviortype determining module 43 is configured to determine the medication behavior type of the target patient in the n+1th time period according to the probability sequence of medication behavior;

确定发送模块44，用于根据上述用药行为类型和上述非用药行为数据确定目标干预策略，并基于上述目标干预策略向上述目标患者发送用药提醒信息。Thedetermination sending module 44 is configured to determine a target intervention strategy according to the above-mentioned medication behavior type and the above-mentioned non-medication behavior data, and send medication reminder information to the above-mentioned target patient based on the above-mentioned target intervention strategy.

在一些可行的实施方式中，上述目标干预策略中包括用药提醒方式和用药提醒频率；In some feasible implementation manners, the above-mentioned target intervention strategy includes medication reminder mode and medication reminder frequency;

上述确定发送模块44用于在预设时间段内，按照上述用药提醒频率以上述用药提醒方式向上述目标患者发送上述用药提醒信息；The aforementioneddetermination sending module 44 is configured to send the aforementioned medication reminder information to the aforementioned target patient in the aforementioned medication reminder manner according to the aforementioned medication reminder frequency within a preset time period;

其中，上述用药提醒方式包括文字提示或语音提示。Wherein, the above-mentioned medication reminding method includes text prompt or voice prompt.

在一些可行的实施方式中，上述概率序列学习模块42用于：In some feasible implementation manners, the aforementioned probability sequence learning module 42 is used to:

将上述预设时长内的行为数据按照预设顺序进行排序以得到排序后的行为数据序列，将上述行为数据序列输入上述LSTM；Sorting the behavior data within the aforementioned preset duration according to a preset order to obtain a sorted behavior data sequence, and inputting the aforementioned behavior data sequence into the aforementioned LSTM;

通过上述LSTM以多种用药行为类型的多分类问题为学习任务来学习上述n个时间段中各个时间段的行为数据对应的下一时间段的用药行为类型。Through the above-mentioned LSTM, a multi-classification problem of multiple medication behavior types is used as a learning task to learn the medication behavior type of the next time period corresponding to the behavior data of each time period in the n time periods.

在一些可行的实施方式中，上述用药行为概率序列中一个用药行为概率对应一个用药行为类型；In some feasible implementation manners, one medication behavior probability corresponds to one medication behavior type in the above-mentioned medication behavior probability sequence;

上述行为类型确定模块43，用于将上述用药行为概率序列中的最大用药行为概率所对应的用药行为类型确定为上述目标患者在上述第n+1时间段的用药行为类型。The behaviortype determining module 43 is configured to determine the medication behavior type corresponding to the maximum medication behavior probability in the medication behavior probability sequence as the medication behavior type of the target patient in the n+1 time period.

在一些可行的实施方式中，上述确定发送模块44，包括：In some feasible implementation manners, the foregoing determining and sendingmodule 44 includes:

标签生成单元441，用于根据上述用药行为类型和上述非用药行为数据生成上述目标患者的自我管理标签；Thelabel generating unit 441 is configured to generate the self-management label of the target patient according to the above-mentioned medication behavior type and the above-mentioned non-medication behavior data;

匹配度计算单元442，用于计算上述自我管理标签与干预策略集合中各个干预策略的策略标签的匹配度以得到多个匹配度值，上述预设干预策略集合包括多个干预策略，一个干预策略携带至少一个策略标签；The matching degree calculation unit 442 is configured to calculate the matching degree between the self-management label and the strategy label of each intervention strategy in the intervention strategy set to obtain multiple matching degree values. The preset intervention strategy set includes multiple intervention strategies, one intervention strategy Carry at least one policy tag;

策略确定单元443，用于将上述多个匹配度值中最大值对应的干预策略确定为上述目标干预策略。The strategy determining unit 443 is configured to determine the intervention strategy corresponding to the maximum value among the multiple matching degree values as the target intervention strategy.

在一些可行的实施方式中，上述行为数据获取模块41，包括：In some feasible implementation manners, the aforementioned behaviordata acquisition module 41 includes:

发送提示单元411，用于按照预设频率向目标患者发送行为数据采集提示以提示上述目标患者按照目标数据采集形式反馈行为数据，其中，上述目标数据采集形式包括网页链接采集或二维码采集；The sendingprompt unit 411 is configured to send behavior data collection prompts to the target patient at a preset frequency to prompt the target patient to feed back behavior data according to the target data collection form, where the target data collection form includes web link collection or QR code collection;

接收数据单元412，用于接收上述目标患者反馈的行为数据以作为上述目标患者在预设时长内的行为数据。Thedata receiving unit 412 is configured to receive the behavior data fed back by the target patient as the behavior data of the target patient within a preset time period.

可以理解的，该患者用药行为干预装置4用于实现图2和图3实施例中用药行为干预平台所执行的步骤。关于图4的患者用药行为干预装置4包括的功能块的具体实现方式及相应的有益效果，可参考前述图2和图3的实施例的具体介绍，这里不赘述。It is understandable that the patient medicationbehavior intervention device 4 is used to implement the steps performed by the medication behavior intervention platform in the embodiment of FIG. 2 and FIG. 3. For the specific implementation manners and corresponding beneficial effects of the functional blocks included in the patient medicationbehavior intervention device 4 in FIG. 4, reference may be made to the specific introduction of the foregoing embodiments in FIG. 2 and FIG. 3, which will not be repeated here.

上述图4所示实施例中的患者用药行为干预装置4可以以图5所示的服务器500来实现。请参见图5，是本申请提供的一种服务器的结构示意图。如图5所示，上述服务器500可以包括：一个或多个处理器501、存储器502和收发器503。上述处理器501、存储器502和收发器503相互连接，如通过总线504连接。其中，上述收发器503用于接收或者发送数据，存储器502用于存储程序代码(或者可称为程序指令)，处理器501用于调用所述程序代码执行上述方法。可选的，上述存储器502用于存储计算机程序，该计算机程序包括程序指令；处理器501用于执行存储器502存储的程序指令，执行如下操作：The patient medicationbehavior intervention device 4 in the embodiment shown in FIG. 4 may be implemented by the server 500 shown in FIG. 5. Please refer to FIG. 5, which is a schematic diagram of the structure of a server provided in this application. As shown in FIG. 5, the foregoing server 500 may include: one ormore processors 501, amemory 502, and atransceiver 503. Theaforementioned processor 501,memory 502, andtransceiver 503 are connected to each other, such as by abus 504. The above-mentionedtransceiver 503 is used to receive or send data, thememory 502 is used to store program codes (or may be called program instructions), and theprocessor 501 is used to call the program codes to execute the above-mentioned methods. Optionally, the foregoingmemory 502 is configured to store a computer program, and the computer program includes program instructions; theprocessor 501 is configured to execute the program instructions stored in thememory 502, and perform the following operations:

在一些可行的实施方式中，上述处理器501通过长短时记忆网络LSTM对上述行为数据进行学习，具体执行以下操作：In some feasible implementation manners, the foregoingprocessor 501 learns the foregoing behavior data through a long and short-term memory network LSTM, and specifically performs the following operations:

上述处理器501根据上述用药行为概率序列确定上述目标患者在上述第n+1时间段的用药行为类型，具体执行以下操作：Theprocessor 501 determines the medication behavior type of the target patient in the n+1th time period according to the medication behavior probability sequence, and specifically performs the following operations:

将上述用药行为概率序列中的最大用药行为概率所对应的用药行为类型确定为上述目标患者在上述第n+1时间段的用药行为类型。Determine the medication behavior type corresponding to the maximum medication behavior probability in the medication behavior probability sequence as the medication behavior type of the target patient in the n+1th time period.

在一些可行的实施方式中，上述处理器501根据上述用药行为类型和上述非用药行为数据确定目标干预策略，具体执行以下操作：In some feasible implementation manners, theaforementioned processor 501 determines a target intervention strategy according to the aforementioned medication behavior type and the aforementioned non- medication behavior data, and specifically executes the following operations:

根据上述用药行为类型和上述非用药行为数据生成上述目标患者的自我管理标签；Generate the self-management label of the target patient according to the above-mentioned medication behavior type and the above-mentioned non-medication behavior data;

计算上述自我管理标签与干预策略集合中各个干预策略的策略标签的匹配度以得到多个匹配度值，上述预设干预策略集合包括多个干预策略，一个干预策略携带至少一个策略标签；Calculating the matching degree between the self-management label and the strategy label of each intervention strategy in the intervention strategy set to obtain multiple matching degree values, the preset intervention strategy set includes a plurality of intervention strategies, and one intervention strategy carries at least one strategy label;

将上述多个匹配度值中最大值对应的干预策略确定为上述目标干预策略。The intervention strategy corresponding to the maximum value among the multiple matching degree values is determined as the target intervention strategy.

上述处理器501基于上述目标干预策略向上述目标患者发送用药提醒信息，具体执行以下操作：Theprocessor 501 sends medication reminder information to the target patient based on the target intervention strategy, and specifically performs the following operations:

在预设时间段内，按照上述用药提醒频率以上述用药提醒方式向上述目标患者发送上述用药提醒信息；Within a preset time period, send the above-mentioned medication reminder information to the above-mentioned target patient in the above-mentioned medication reminder manner according to the above-mentioned medication reminder frequency;

在一些可行的实施方式中，上述处理器501获取目标患者在预设时长内的行为数据，具体执行以下操作：In some feasible implementation manners, theaforementioned processor 501 obtains the behavior data of the target patient within a preset period of time, and specifically executes the following operations:

按照预设频率向目标患者发送行为数据采集提示以提示上述目标患者按照目标数据采集形式反馈行为数据，其中，上述目标数据采集形式包括网页链接采集或二维码采集；Send behavior data collection prompts to the target patient at a preset frequency to prompt the target patient to feed back behavior data in a target data collection form, where the target data collection form includes web link collection or QR code collection;

接收上述目标患者反馈的行为数据以作为上述目标患者在预设时长内的行为数据。The behavior data fed back by the target patient is received as the behavior data of the target patient within a preset time period.

此外，这里需要指出的是：本申请还提供了一种计算机可读存储介质，且所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行以实现上述方法。可选的，上述计算机可读存储介质中存储有前文提及的患者用药行为干预装置4所执行的计算机程序，且上述计算机程序包括程序指令，当上述处理器执行上述程序指令时，能够执行前文图2或图3对应实施例中对上述患者用药行为干预方法的描述，因此，这里将不再进行赘述。另外，对采用相同方法的有益效果描述，也不再进行赘述。对于本申请所涉及的计算机可读存储介质实施例中未披露的技术细节，请参照本申请方法实施例的描述。作为示例，程序指令可被部署为在一个计算设备上执行，或者在位于一个地点的多个计算设备上执行，又或者，在分布在多个地点且通过通信网络互连的多个计算设备上执行，分布在多个地点且通过通信网络互连的多个计算设备可以组成区块链系统。In addition, it should be pointed out here that this application also provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement the above method. Optionally, the aforementioned computer-readable storage medium stores the aforementioned computer program executed by the patient medicationbehavior intervention device 4, and the aforementioned computer program includes program instructions. When the aforementioned processor executes the aforementioned program instructions, the aforementioned program instructions can be executed. Fig. 2 or Fig. 3 corresponds to the description of the above-mentioned patient medication behavior intervention method in the embodiment, therefore, it will not be repeated here. In addition, the description of the beneficial effects of using the same method will not be repeated. For technical details that are not disclosed in the embodiment of the computer-readable storage medium involved in this application, please refer to the description of the method embodiment of this application. As an example, the program instructions can be deployed to be executed on one computing device, or on multiple computing devices located in one location, or on multiple computing devices that are distributed in multiple locations and interconnected by a communication network Implementation, multiple computing devices distributed in multiple locations and interconnected through a communication network can form a blockchain system.

可选的，本申请涉及的存储介质如计算机可读存储介质可以是非易失性的，也可以是易失性的。Optionally, the storage medium involved in this application, such as a computer-readable storage medium, may be non-volatile or volatile.

本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，上述的程序可存储于一计算机可读取存储介质中，该程序在执行时，可包括如上述各方法的实施例的流程。其中，上述的存储介质可为磁碟、光盘、只读存储记忆体(Read-Only Memory，ROM)或随机存储记忆体(Random Access Memory，RAM)等。A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing relevant hardware through a computer program. The above-mentioned program can be stored in a computer readable storage medium. When executed, it may include the procedures of the above-mentioned method embodiments. Among them, the aforementioned storage medium may be a magnetic disk, an optical disc, a read-only memory (Read-Only Memory, ROM), or a random access memory (Random Access Memory, RAM), etc.

本申请提供的方法及相关装置是参照本申请提供的方法流程图和/或结构示意图来描述的，具体可由计算机程序指令实现方法流程图和/或结构示意图的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上，使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理，从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或结构示意一个方框或多个方框中指定的功能的步骤。The methods and related devices provided in this application are described with reference to the method flowcharts and/or structural schematic diagrams provided in this application. Specifically, each process and/or block of the method flowcharts and/or structural schematic diagrams can be implemented by computer program instructions. And a combination of processes and/or blocks in flowcharts and/or block diagrams. These computer program instructions can also be loaded on a computer or other programmable data processing equipment, so that a series of operation steps are executed on the computer or other programmable equipment to produce computer-implemented processing, so as to execute on the computer or other programmable equipment. The instructions provide steps for implementing the functions specified in one block or multiple blocks in the flow chart or the flow chart and/or the structure.

以上所揭露的仅为本申请较佳实施例而已，当然不能以此来限定本申请之权利范围，因此依本申请权利要求所作的等同变化，仍属本申请所涵盖的范围。The above-disclosed are only preferred embodiments of this application, and of course the scope of rights of this application cannot be limited by this. Therefore, equivalent changes made in accordance with the claims of this application still fall within the scope of this application.