WO2021189958A1 - 慢病随访记录收集方法、装置、设备及存储介质 - Google Patents

慢病随访记录收集方法、装置、设备及存储介质 Download PDF

Info

Publication number
WO2021189958A1
WO2021189958A1 PCT/CN2020/135272 CN2020135272W WO2021189958A1 WO 2021189958 A1 WO2021189958 A1 WO 2021189958A1 CN 2020135272 W CN2020135272 W CN 2020135272W WO 2021189958 A1 WO2021189958 A1 WO 2021189958A1
Authority
WO
WIPO (PCT)
Prior art keywords
follow
chronic disease
data
patient
abnormal
Prior art date
Application number
PCT/CN2020/135272
Other languages
English (en)
French (fr)
Inventor
左磊
赵惟
徐卓扬
孙行智
胡岗
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2021189958A1 publication Critical patent/WO2021189958A1/zh

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/60—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for patient-specific data, e.g. for electronic patient records
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/70—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for mining of medical data, e.g. analysing previous cases of other patients
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
    • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Definitions

  • This application relates to the technical field of medical big data, and in particular to a method, device, equipment, and storage medium for collecting chronic disease follow-up records.
  • the inventor realizes that the data collected by these collection methods rely heavily on the subjective answers, filling in, and the recording habits of medical staff of the enrolled patients. The quality of the data is interfered by many subjective and external factors. Once collected, the data will be analyzed later. In the process, it is difficult to correct the data.
  • a method for collecting follow-up records of chronic diseases comprising:
  • the follow-up record collection is carried out according to the trained long and short-term memory network attention model, and the follow-up record collection results are obtained.
  • the chronic disease follow-up record collection device includes:
  • a collection generation module configured to generate a chronic disease content collection according to the chronic disease data, and obtain a chronic disease patient feature collection according to the chronic disease content collection;
  • An abnormality acquisition module for acquiring characteristic abnormal conditions corresponding to different patient characteristics in the chronic disease patient characteristic set
  • the model training module is used to obtain historical follow-up data within a preset time, and train a long- and short-term memory network attention model according to the historical follow-up data and the characteristic abnormal conditions;
  • the record collection module is used to collect follow-up records according to the trained long- and short-term memory network attention model, and obtain follow-up record collection results.
  • a chronic disease follow-up record collection device comprising: a memory, a processor, and a chronic disease follow-up record collection program stored on the memory and runable on the processor, the chronic disease follow-up record collection program When executed by the processor, the following steps are implemented:
  • the follow-up record collection is carried out according to the trained long and short-term memory network attention model, and the follow-up record collection results are obtained.
  • a storage medium storing a chronic disease follow-up record collection program, and the following steps are implemented when the chronic disease follow-up record collection program is executed by a processor:
  • the follow-up record collection is carried out according to the trained long and short-term memory network attention model, and the follow-up record collection results are obtained.
  • This application facilitates the staff to confirm information to the patient in time and improve the quality of the database.
  • FIG. 1 is a schematic structural diagram of a chronic disease follow-up record collection device in a hardware operating environment involved in a solution of an embodiment of the present application;
  • FIG. 2 is a schematic flow chart of the first embodiment of the method for collecting follow-up records of chronic diseases according to the application;
  • FIG. 3 is a schematic flowchart of a second embodiment of a method for collecting follow-up records of chronic diseases according to the application;
  • FIG. 4 is a schematic flowchart of a third embodiment of a method for collecting follow-up records of chronic diseases according to the application;
  • Fig. 5 is a structural block diagram of the first embodiment of the chronic disease follow-up record collection device according to the present application.
  • FIG. 1 is a schematic structural diagram of a chronic disease follow-up record collection device in a hardware operating environment involved in a solution of an embodiment of the application.
  • the chronic disease follow-up record collection device may include a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005.
  • the communication bus 1002 is used to implement connection and communication between these components.
  • the user interface 1003 may include a display screen (Display) and an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface.
  • the network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WI-FIdelity, WI-FI) interface).
  • WI-FIdelity wireless fidelity
  • the memory 1005 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory.
  • RAM Random Access Memory
  • NVM Non-Volatile Memory
  • the memory 1005 may also be a storage device independent of the aforementioned processor 1001.
  • FIG. 1 does not constitute a limitation on the chronic disease follow-up record collection device, and may include more or less components than shown, or a combination of certain components, or different components Layout.
  • the memory 1005 as a storage medium may include an operating system, a data storage module, a network communication module, a user interface module, and a chronic disease follow-up record collection program.
  • the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with users; processing in the chronic disease follow-up record collection device of this application
  • the device 1001 and the memory 1005 may be set in a chronic disease follow-up record collection device, which calls the chronic disease follow-up record collection program stored in the memory 1005 through the processor 1001, and executes the following steps:
  • the follow-up record collection is carried out according to the trained long and short-term memory network attention model, and the follow-up record collection results are obtained.
  • the embodiment of the present application provides a method for collecting follow-up records of chronic diseases.
  • FIG. 2 is a schematic flowchart of the first embodiment of the method for collecting follow-up records of chronic diseases according to this application.
  • the chronic disease follow-up record collection method includes the following steps:
  • Step S10 Generate a chronic disease content set according to the chronic disease data, and obtain a chronic disease patient feature set according to the chronic disease content set.
  • the chronic disease data are various types of chronic disease data obtained in combination with clinical guidelines, expert consensus, relevant literature, and clinical experience.
  • atrial fibrillation that is, atrial fibrillation, which is the most common persistent arrhythmia
  • first obtain the disease information of the target patient for the current chronic disease follow-up and learn that the target patient suffers from the disease information.
  • For atrial fibrillation collect chronic disease data such as clinical guidelines, expert consensus, relevant literature, and clinical experience related to atrial fibrillation, and obtain characteristics related to atrial fibrillation according to the chronic disease data to generate a feature set of chronic disease patients.
  • step S10 specifically includes: acquiring chronic disease data collected during the chronic disease follow-up process, and generating a chronic disease follow-up table and a chronic disease baseline table according to the patient chronic disease data;
  • the chronic disease follow-up table and the chronic disease baseline table generate a chronic disease content collection; and
  • a chronic disease patient characteristic collection is obtained from the chronic disease content collection according to the chronic disease clinical demand information.
  • the chronic disease baseline table is the basic items that need to be recorded for the patient during the chronic disease process, such as: personal information such as the patient's name and age, the patient's hospital stay, the patient's length of illness, symptoms, treatment status, disease history, family history, examination Inspection;
  • the follow-up record table contains various items that need to be obtained through follow-up, such as: current follow-up time, current symptoms, current treatment conditions, current curative effects, clinical events, inspections and so on.
  • the collection of items in the chronic disease baseline table and the chronic disease follow-up table is the chronic disease content collection.
  • the chronic disease patient feature set is a subset of the chronic disease content set.
  • the chronic disease clinical demand information is a patient characteristic that needs to be focused clinically, and the clinical demand can be obtained through chronic disease data. For example: Taking atrial fibrillation as an example, clinically, it is necessary to focus on the anticoagulation of patients with atrial fibrillation to prevent adverse events such as stroke.
  • the patient characteristics that need to be focused mainly include: whether to take anticoagulants Drugs, what kind of anticoagulant (warfarin/new anticoagulant), drug dosage, whether there is a stroke, whether there is a major hemorrhage (especially pay attention to whether there is an intracranial hemorrhage), if a stroke has occurred Risk and so on. Therefore, the elements contained in the feature set of patients with chronic diseases may be: the type of medicine taken by the patient, the amount of medicine, the patient's history of illness events, the patient's onset risk events, and the like.
  • Step S20 Obtain characteristic abnormal conditions corresponding to different patient characteristics in the chronic disease patient characteristic set.
  • the first type is data missing abnormalities, for example: a certain item is required, but the patient or follow-up person does not fill in; the second type is abnormal data beyond the normal range
  • the drug dosage is generally 1 to 3 tablets a day. If a patient fills in to take 10 tablets a day or 0 tablets a day, it is defined as an abnormal value; the third category is a logical error, for example, the doctor’s prescription does not indicate the patient Taking a certain drug, the patient has a record of taking the drug; the patient has not taken a certain drug, but the dose of the drug has a specific value.
  • step S20 specifically includes: determining the characteristic relationship between the characteristics of each patient in the chronic disease patient characteristic set according to the chronic disease baseline table and the chronic disease follow-up table; The relationship is input into the preset rule engine to obtain the characteristic logic error condition; the characteristic preset range condition is determined according to the chronic disease baseline table and the chronic disease follow-up table; the characteristic logic error condition and the characteristic preset range condition are used as Characteristic abnormal conditions.
  • the characteristic relationship for example, if a patient takes a certain drug every day, a corresponding change in the condition occurs; the change in the condition is associated with the amount of drug taken.
  • the certain medicine and the other medicine are incompatible with each other, and cannot be taken at the same time, then there is a mutually exclusive relationship between the dosage of the certain medicine and the other medicine.
  • the feature relationship is converted into a language understood by the computer, and the converted feature relationship is input into the rule engine to obtain various feature logic errors.
  • the characteristic logic error generates a characteristic logic error condition. If the features are mutually exclusive, the features do not satisfy the feature logic error condition.
  • the feature preset range condition is the preset range of the value of the data corresponding to the feature. If the data is missing and the data exceeds the normal value range, the feature does not meet the feature preset range condition and the feature is abnormal.
  • Step S30 Obtain historical follow-up data within a preset time, and train a long- and short-term memory network attention model according to the historical follow-up data and the characteristic abnormal condition.
  • the preset time may be one year or six months
  • the historical follow-up data of the target patient within one year is acquired
  • the long-term short-term memory network attention model is trained according to the historical follow-up data and the characteristic abnormal conditions to A long-term short-term memory network attention model that meets the needs of abnormal detection in follow-up records.
  • the long and short-term memory network attention model includes an input layer Input, an embedding layer Embedding, a long short-term memory network (Long short-term memory, LSTM) layer, an attention mechanism (Attention) layer, and a counterattack function layer Sigmoid And the output layer Output.
  • step S30 specifically includes: obtaining historical follow-up data within a preset time according to the follow-up record table; constructing the long- and short-term memory network attention model according to the characteristic abnormal condition; The historical follow-up data is input into the long- and short-term memory network attention model at a time point corresponding to the historical follow-up data for model training.
  • obtaining the patient's historical follow-up data within one year including at least one chronic disease baseline table data and two follow-up record data.
  • Each follow-up record data is used as the input of a time point of the attention model of the long and short-term memory network, and each input is an N-dimensional vector, and N is the number of features of patients with chronic diseases.
  • the long-short-term memory network attention model outputs an M-dimensional vector, and M is the probability of the output feature, representing the probability of occurrence of M abnormal conditions or adverse events. According to the output value of the attention mechanism, it is possible to see which feature at which time point has a greater contribution to the output result, and it is easier to know the patient's physical condition in the future to monitor and further improve.
  • Step S40 Perform follow-up record collection according to the trained long and short-term memory network attention model, and obtain follow-up record collection results.
  • step S40 specifically includes: obtaining current follow-up data in real time according to the follow-up record table, and inputting the current follow-up data into the trained long and short-term memory network attention model;
  • the output result of the long-short-term memory network attention model detects whether there are abnormal parameters in the current follow-up data; when the abnormal parameters are abnormal parameters, the historical follow-up data of the patients corresponding to the abnormal conditions are acquired, and according to the current
  • the follow-up data and the historical follow-up data are used to analyze the patient's condition; the disease analysis result and the current follow-up data are used as the follow-up record collection result.
  • the current follow-up data is acquired according to the follow-up record table, the patient's chronic disease condition is checked in real time, and the current follow-up data is input into the trained long and short-term memory network attention model.
  • the output result is the probability of occurrence of multiple abnormal conditions or adverse events, corresponding to multiple patient characteristics one by one.
  • the abnormal condition parameter is the occurrence probability of the patient at various clinical times in the future. For example, taking atrial fibrillation as an example, it may be an embolic event, a bleeding event, or a death event.
  • a certain abnormal parameter is extremely high, that is, when the probability of a certain event is greater than the preset probability, it is judged that the feature corresponding to the event is abnormal.
  • the patient characteristic may be the value corresponding to the transaminase, the patient's transaminase value rises and is higher than the preset value, the patient is at risk of complications such as liver cirrhosis, and it is estimated that the patient will develop liver cirrhosis through the model
  • the probability of the risk is greater than the preset probability, and medical staff can prescribe the right medicine to the patient based on the transaminase value to prevent complications or deterioration of the condition.
  • the long and short-term memory network attention model is trained through historical follow-up data, so that the long- and short-term memory network attention model can meet the requirements for collecting follow-up data, and the trained long- and short-term memory network attention model is used for follow-up data collection.
  • the patient’s condition be evaluated and predicted based on the patient’s each follow-up data, but also the risk factors that have a greater impact on the condition can be identified, which is convenient for patients to grasp their own physical condition in time and make improvements to the risk factors, which is more conducive to personality Prevent the deterioration of the disease and the occurrence of complications chemically. It is convenient for the staff to confirm the information to the patient in time and improve the quality of the database.
  • FIG. 3 is a schematic flowchart of a second embodiment of a method for collecting follow-up records of chronic diseases according to this application.
  • step S40 after the step of detecting whether the current follow-up data has abnormal parameters according to the output result of the trained long-short-term memory network attention model in step S40, include:
  • Step S401 When the abnormal parameter is a data abnormal parameter, obtain patient information corresponding to the data abnormal parameter.
  • the data abnormal parameters include: the first type is data missing abnormalities, and the second type is abnormal values beyond the normal range. In this case, there is an erroneous record of the data, and the corresponding patient needs to be found to re-acquire the data. When the corresponding patient information is obtained, the patient information is displayed to the corresponding medical staff or the patient himself.
  • Step S402 Receive corrected follow-up data based on the patient information feedback, and replace the original data corresponding to the abnormal data with the corrected follow-up data to generate corrected current follow-up data.
  • Step S403 Input the corrected current follow-up data into the trained long and short-term memory network attention model.
  • the corrected current follow-up data is input into the trained long and short-term memory network attention model, based on the first embodiment, the function of disease analysis is performed, and the current follow-up data and the result of the disease analysis are taken as The follow-up record collection results.
  • This embodiment uses the trained long and short-term memory network attention model to check the erroneous data in the follow-up data, and promptly notify the relevant personnel to correct the erroneous data, which helps to reduce the rate of misdiagnosis and missed diagnosis, and the condition is based on the corrected data.
  • Predictive analysis can help improve the accuracy of disease predictive analysis and assist medical personnel in obtaining more comprehensive follow-up data.
  • FIG. 4 is a schematic flowchart of a third embodiment of a method for collecting follow-up records of chronic diseases according to this application.
  • the abnormal parameter is an abnormal condition parameter in the step S40
  • the historical follow-up data of the patient corresponding to the abnormal condition is acquired, and the current follow-up data is compared with all the patients.
  • Step S404 when the abnormal parameter is an abnormal condition parameter, obtain historical follow-up data of the patient corresponding to the abnormal condition parameter.
  • Step S405 Acquire the patient's condition change degree according to the preset condition change information and the historical follow-up data.
  • the degree of change in the condition of the patient refers to which stage of the patient's condition belongs to.
  • the preset disease condition change information level is the common pathological stage of the target chronic disease obtained according to the chronic disease data
  • the patient’s current condition status is obtained according to the historical follow-up data
  • the condition status is determined according to the preset condition change information What degree of disease change. For example: Taking the common pathological changes of fatty liver as an example, it can be divided into mild fatty liver, moderate and severe.
  • the acquired historical follow-up data is that the patient’s B-ultrasound shows that the liver has changed from no fat accumulation to a large amount of fat accumulation, and the disease state is a large amount of fat accumulation, and it can be judged that the patient's current condition change degree is severe fatty liver.
  • Step S406 Predict the condition according to the current follow-up data and the degree of change of the condition, and use the condition prediction result as the condition analysis result.
  • condition prediction result is used as the condition analysis result.
  • the embodiment of the present application also proposes a storage medium.
  • the storage medium may be volatile or non-volatile, and a chronic disease follow-up record collection program is stored on the storage medium, and the chronic disease follow-up record When the collection program is executed by the processor, the following steps are implemented:
  • the follow-up record collection is carried out according to the trained long and short-term memory network attention model, and the follow-up record collection results are obtained.
  • Fig. 5 is a structural block diagram of a first embodiment of a chronic disease follow-up record collection device according to the present application.
  • the chronic disease follow-up record collection device proposed in the embodiment of the present application includes:
  • the collection generating module 501 is configured to generate a chronic disease content collection according to the chronic disease data, and obtain a chronic disease patient feature collection according to the chronic disease content collection;
  • An abnormality acquisition module 502 configured to acquire characteristic abnormal conditions corresponding to different patient characteristics in the chronic disease patient characteristic set;
  • the model training module 503 is configured to obtain historical follow-up data within a preset time, and train a long- and short-term memory network attention model according to the historical follow-up data and the characteristic abnormal conditions;
  • the record collection module 504 is configured to collect follow-up records according to the trained long- and short-term memory network attention model, and obtain follow-up record collection results.
  • the long and short-term memory network attention model is trained through historical follow-up data, so that the long- and short-term memory network attention model can meet the requirements for collecting follow-up data, and the trained long- and short-term memory network attention model is used for follow-up data collection.
  • the patient’s condition be evaluated and predicted based on the patient’s each follow-up data, but also the risk factors that have a greater impact on the condition can be identified, which is convenient for patients to grasp their own physical condition in time and make improvements to the risk factors, which is more conducive to personality Prevent the deterioration of the disease and the occurrence of complications chemically. It is convenient for the staff to confirm the information to the patient in time and improve the quality of the database.
  • the set generation module 501 is also used to obtain the chronic disease data of the patient collected during the chronic disease follow-up process, and generate the chronic disease follow-up table and the chronic disease baseline table according to the chronic disease data of the patient; according to the chronic disease follow-up table and the chronic disease
  • the disease baseline table generates a chronic disease content collection; according to the chronic disease clinical demand information, a chronic disease patient characteristic collection is obtained from the chronic disease content collection.
  • the abnormality acquisition module 502 is further configured to determine the characteristic relationship between the characteristics of each patient in the chronic disease patient characteristic set according to the chronic disease baseline table and the chronic disease follow-up table; input the characteristic relationship into the predictive Set up a rule engine to obtain characteristic logic error conditions; determine characteristic preset range conditions according to the chronic disease baseline table and the chronic disease follow-up table; use the characteristic logic error conditions and the characteristic preset range conditions as characteristic abnormal conditions .
  • model training module 503 is also used to obtain historical follow-up data within a preset time according to the follow-up record table; construct a long- and short-term memory network attention model according to the characteristic abnormal conditions; and according to the time point corresponding to the historical follow-up data
  • the historical follow-up data is input into the long and short-term memory network attention model for model training.
  • the record collection module 504 is further configured to obtain current follow-up data in real time according to the follow-up record table, and input the current follow-up data into the trained long- and short-term memory network attention model;
  • the output result of the memory network attention model detects whether there are abnormal parameters in the current follow-up data; when the abnormal parameter is an abnormal parameter, the historical follow-up data of the patient corresponding to the abnormal condition is obtained, and the current follow-up data is compared with The historical follow-up data is used to analyze the patient's condition; the disease analysis result and the current follow-up data are used as the follow-up record collection result.
  • the record collection module 504 is further configured to obtain patient information corresponding to the abnormal data parameter when the abnormal parameter is a data abnormal parameter; receive corrected follow-up data based on the patient information feedback, and use the corrected follow-up data The data replaces the original data corresponding to the abnormal data to generate corrected current follow-up data; the corrected current follow-up data is input into the trained long-short-term memory network attention model.
  • the record collection module 504 is also used to obtain the historical follow-up data of the patient corresponding to the abnormal condition parameter when the abnormal parameter is the abnormal condition parameter; obtain the patient according to preset change information of the condition and the historical follow-up data The degree of change in the condition of the disease; the disease condition is predicted based on the current follow-up data and the degree of change in the condition, and the result of the disease condition prediction is used as the result of the condition analysis.
  • the chronic disease follow-up record collection method provided by the present application further ensures the privacy and security of all the above-mentioned data
  • all the above-mentioned data can also be stored in a node of a blockchain.
  • the feature collection of patients with chronic diseases and the collection results of follow-up records, etc., these data can be stored in the blockchain node.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information for verification. The validity of the information (anti-counterfeiting) and the generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
  • the technical solution of this application essentially or the part that contributes to the existing technology can be embodied in the form of a software product, and the computer software product is stored in a storage medium (such as read-only memory/random access
  • the memory, magnetic disk, optical disk includes several instructions to make a terminal device (which can be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) execute the chronic disease follow-up record collection described in each embodiment of this application method.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Public Health (AREA)
  • Medical Informatics (AREA)
  • Primary Health Care (AREA)
  • Epidemiology (AREA)
  • Databases & Information Systems (AREA)
  • Pathology (AREA)
  • Measuring And Recording Apparatus For Diagnosis (AREA)
  • Medical Treatment And Welfare Office Work (AREA)

Abstract

一种慢病随访记录收集方法、装置、设备及存储介质，该方法包括：根据慢病数据生成慢病内容集合，并根据慢病内容集合获取慢病患者特征集合（S10）；获取慢病患者特征集合中不同患者特征对应的特征异常条件（S20）；获取预设时间内的历史随访数据，根据历史随访数据与特征异常条件训练长短期记忆网络注意力模型（S30）；根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果（S40）。通过历史随访数据对长短期记忆网络注意力模型进行训练，使用训练后的模型进行随访数据收集，可以根据随访数据对患者病情进行评估和预测，便于工作人员及时确认随访信息，提高数据库的质量。

Description

慢病随访记录收集方法、装置、设备及存储介质
本申请要求于2020年10月9日提交中国专利局、申请号为CN202011074729.5       ，发明名称为“慢病随访记录收集方法、装置、设备及存储介质”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。
技术领域
本申请涉及医疗大数据技术领域，尤其涉及一种慢病随访记录收集方法、装置、设备及存储介质。
 背景技术
目前，医疗大数据已经被广泛应用于临床决策支持、药物研发等方面，但与此同时，其数据质量不高、结构化和标准化不足等问题也限制了进一步的发展和应用。数据质量不高的一个重要原因是发生在数据收集过程中。以慢病随访记录收集为例，目前常见的方法是电话随访：护士以电话形式回访入组患者，按照预先设定好的问题对患者进行依次询问，患者再根据自身实际情况进行回复，结果由护士进行记录；问卷随访：入组患者在线填写随访表，数据结果由工作人员进行整理和结构化储存。
 技术问题
发明人意识到这些收集方法收集到的数据严重依赖于入组患者的主观回答、填写以及医护工作人员的记录习惯，数据质量受很多主观因素和外界因素的干扰，且一旦收集，后期在数据分析过程中，很难再对数据进行校正。
上述内容仅用于辅助理解本申请的技术方案，并不代表承认上述内容是现有技术。
 技术解决方案
一种慢病随访记录收集方法，所述方法包括：
根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
一种慢病随访记录收集装置所述慢病随访记录收集装置包括：
集合生成模块，用于根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
异常获取模块，用于获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
模型训练模块，用于获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
记录收集模块，用于根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
一种慢病随访记录收集设备，所述设备包括：存储器、处理器及存储在所述存储器上并可在所述处理器上运行的慢病随访记录收集程序，所述慢病随访记录收集程序被处理器执行时实现如下步骤：
根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
一种存储介质，所述存储介质上存储有慢病随访记录收集程序，所述慢病随访记录收集程序被处理器执行时实现如下步骤：
根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
本申请便于工作人员及时的向患者确认信息，提高数据库的质量。
 附图说明
图1是本申请实施例方案涉及的硬件运行环境的慢病随访记录收集设备的结构示意图；
图2为本申请慢病随访记录收集方法第一实施例的流程示意图；
图3为本申请慢病随访记录收集方法第二实施例的流程示意图；
图4为本申请慢病随访记录收集方法第三实施例的流程示意图；
图5为本申请慢病随访记录收集装置第一实施例的结构框图。
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
 本发明的实施方式
应当理解，此处所描述的具体实施例仅用以解释本申请，并不用于限定本申请。
参照图1，图1为本申请实施例方案涉及的硬件运行环境的慢病随访记录收集设备结构示意图。
如图1所示，该慢病随访记录收集设备可以包括：处理器1001，例如中央处理器（Central Processing Unit，CPU），通信总线1002、用户接口1003，网络接口1004，存储器1005。其中，通信总线1002用于实现这些组件之间的连接通信。用户接口1003可以包括显示屏（Display）、输入单元比如键盘（Keyboard），可选用户接口1003还可以包括标准的有线接口、无线接口。网络接口1004可选的可以包括标准的有线接口、无线接口（如无线保真（WIreless-FIdelity，WI-FI）接口）。存储器1005可以是高速的随机存取存储器（Random Access Memory，RAM）存储器，也可以是稳定的非易失性存储器（Non-Volatile Memory，NVM），例如磁盘存储器。存储器1005可选的还可以是独立于前述处理器1001的存储装置。
本领域技术人员可以理解，图1中示出的结构并不构成对慢病随访记录收集设备的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置。
如图1所示，作为一种存储介质的存储器1005中可以包括操作系统、数据存储模块、网络通信模块、用户接口模块以及慢病随访记录收集程序。
在图1所示的慢病随访记录收集设备中，网络接口1004主要用于与网络服务器进行数据通信；用户接口1003主要用于与用户进行数据交互；本申请慢病随访记录收集设备中的处理器1001、存储器1005可以设置在慢病随访记录收集设备中，所述慢病随访记录收集设备通过处理器1001调用存储器1005中存储的慢病随访记录收集程序，并执行如下步骤：
根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
本申请慢病随访记录收集设备的其他实施例或具体实现方式可参照下述各方法实施例，此处不再赘述。
本申请实施例提供了一种慢病随访记录收集方法，参照图2，图2为本申请慢病随访记录收集方法第一实施例的流程示意图。
本实施例中，所述慢病随访记录收集方法包括以下步骤：
步骤S10：根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合。
需要说明的是，所述慢病数据为结合临床指南、专家共识、相关文献、临床经验获取的各类慢病数据。本实施例中以房颤（即心房颤动，是最常见的持续性心律失常）为例，例如：首先获取当前慢病随访的目标患者的病症信息，根据所述病症信息得知目标患者患有房颤，搜集房颤相关的临床指南、专家共识、相关文献及临床经验等慢病数据，根据所述慢病数据获取房颤相关的特征，以生成慢病患者特征集合。
进一步地，为获得慢病患者特征集合，步骤S10具体包括：获取慢病随访过程中收集的患者慢病数据，根据所述患者慢病数据生成慢病随访表及慢病基线表；根据所述慢病随访表与所述慢病基线表生成慢病内容集合；根据慢病临床需求信息从所述慢病内容集合中获取慢病患者特征集合。
易于理解的是，所述慢病数据中对应地包含慢病需要关注的慢病内容，根据所述慢病内容进行聚类分析生成慢病基线表、随访记录表。所述慢病基线表为慢病过程中需要对患者进行记录的基础项目，例如：患者姓名年龄等个人信息、患者住院时间、患者患病时长、症状、治疗情况、疾病史、家族史、检验检查；所述随访记录表中包含各项需要通过随访进行获取的项目，例如：当前随访时间、目前的症状、当前的治疗情况、当前疗效、临床事件、检验检查等。所述慢病基线表与所述慢病随访表中的各项目的集合，即为所述慢病内容集合。
易于理解的是，慢病患者特征集合是慢病内容集合的子集合。所述慢病临床需求信息为临床上需要重点关注的患者特征，所述临床需求可以通过慢病数据进行获取。例如：以房颤为例，临床上需要重点关注房颤患者的抗凝情况，以预防脑卒中等不良事件发生，因此在随访过程中，需要重点关注的患者特征主要包括：是否服用抗凝类药物，服用何种抗凝药物（华法林/新型抗凝药），药物剂量，是否有脑卒中事件发生，是否有大出血事件发生（尤其关注是否有颅内出血事件发生），发生脑卒中事件的风险等等。因此，所述慢病患者特征集合中包含的元素可以为：患者服用的药物种类、药量、患者的历史病情事件、患者发病风险事件等。
步骤S20：获取所述慢病患者特征集合中不同患者特征对应的特征异常条件。
易于理解的是，特征异常通常具有三种情况，第一类是数据缺失异常，例如：某一项目为必填事项，但患者或随访人员未填写；第二类是数据超出正常值范围的异常，如药物剂量一般是一天服用1至3片，如果有患者填入一天服用10片或者一天服用0片，则定义为异常值；第三类是逻辑错误，例如：医师处方中并未指示患者服用某一种药物，患者却存在服用该药物的记录；患者并未服用某一种药物，但该药物的剂量却有具体数值。
进一步地，为获取特征异常条件，步骤S20具体包括：根据所述慢病基线表及所述慢病随访表确定所述慢病患者特征集合中各患者特征之间的特征关系；将所述特征关系输入预设规则引擎以获取特征逻辑错误条件；根据所述慢病基线表与所述慢病随访表确定特征预设范围条件；将所述特征逻辑错误条件、所述特征预设范围条件作为特征异常条件。
需要说明的是，对于所述特征关系，例如：患者每天服用某一药物，则出现对应的病情变化；所述病情变化与药物服用量呈关联关系。患者服用某一药物，所述某一药物与另一药物存在药性相克，不能同时服用，则某一药物与另一药物的服用量存在互斥关系。
易于理解的是，基于慢病基线表和慢病随访表的各项目的设计，将特征关系转换为计算机理解的语言，并将转换后的特征关系输入规则引擎以获取各类特征逻辑错误，根据所述特征逻辑错误生成特征逻辑错误条件。若特征之间存在互斥，则特征不满足所述特征逻辑错误条件。
易于理解的是，特征预设范围条件为特征对应的数据的值的预设范围，若数据缺失、数据超出正常值范围，则特征不满足所述特征预设范围条件，特征存在异常。
步骤S30：获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型。
具体实施中，所述预设时间可以为一年或半年，获取目标患者一年内的历史随访数据，根据所述历史随访数据与所述特征异常条件对长短期记忆网络注意力模型进行训练，以符合随访记录异常检测需求的长短期记忆网络注意力模型。
易于理解的是，所述长短期记忆网络注意力模型包含输入层Input、嵌入层Embedding、长短期记忆网络（Long short-term memory，LSTM）层、注意力机制（Attention）层、反击函数层Sigmoid及输出层Output。
进一步地，为对长短期记忆网络注意力模型进行训练，步骤S30具体包括：根据随访记录表获取预设时间内的历史随访数据；根据所述特征异常条件构建长短期记忆网络注意力模型；根据所述历史随访数据对应的时间点将所述历史随访数据输入所述长短期记忆网络注意力模型中进行模型训练。
具体实施中，例如：获取患者一年内的历史随访数据，至少包含一条慢病基线表数据与两条随访记录数据。将每次的随访记录数据作为所述长短期记忆网络注意力模型的一个时间点的输入，每个输入时一个N维向量，N为慢病患者特征数。所述长短期记忆网络注意力模型输出一个M维向量，M为输出的特征的概率数，代表M个病情异常情况或不良事件发生的概率。根据注意力机制的输出值可以看哪个时间点的哪个特征对输出结果的贡献度比较大，更便于知道患者在日后对进行的身体状况进行监测和进一步改善。
步骤S40：根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
进一步地，为进行随访记录收集，步骤S40具体包括：根据所述随访记录表实时获取当前随访数据，并将所述当前随访数据输入训练后的长短期记忆网络注意力模型；根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数；在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析；将病情分析结果与所述当前随访数据作为随访记录收集结果。
需要说明的是，根据所述随访记录表获取当前随访数据，实时对患者慢病病情进行检查，将所述当前随访数据输入训练后的长短期记忆网络注意力模型。所述输出结果为多个病情异常情况或不良事件发生的概率，一一对应多个患者特征。所述病情异常参数为患者未来多种临床时间的发生概率，例如：以房颤为例，可以为栓塞事件、出血事件、死亡事件。在某一异常参数极高，即，某一事件的概率大于预设概率时，判断该事件对应的特征存在着异常。例如：对于脂肪肝，所述患者特征可以为转氨酶对应的数值，患者的转氨酶数值上升且高于预设数值，患者存在肝硬化等并发症的风险，通过所述模型预估到患者发生肝硬化风险的概率大于预设概率，医疗人员可以根据转氨酶数值对患者对症下药，以防并发症的发生或病情恶化。
易于理解的是，利用新收集到的数据可以对模型性能进行评估，也可以优化模型以达到更优性能。
本实施例通过历史随访数据对长短期记忆网络注意力模型进行训练，使得长短期记忆网络注意力模型能够符合收集随访数据的要求，使用训练后的长短期记忆网络注意力模型进行随访数据收集，不仅可以根据患者每次随访数据对患者病情进行评估和预测，还可以找出对病情影响较大的危险因素，便于患者及时的掌握自身身体状况，并对危险因素做出改善，更有利于个性化地预防病情恶化和并发症的发生。便于工作人员及时的向患者确认信息，提高数据库的质量。
参考图3，图3为本申请慢病随访记录收集方法第二实施例的流程示意图。
基于上述第一实施例，在本实施例中，步骤 S40中的所述根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数的步骤之后，还包括：
步骤S401：在所述异常参数为数据异常参数时，获取所述数据异常参数对应的患者信息。
易于理解的是，数据异常参数包括：第一类是数据缺失异常、第二类是超出正常值范围的异常。在这种情况下，存在对数据的错误记录，需要找出对应的患者进行数据重新获取，获取到对应的患者信息时，将所述患者信息展示给对应的医疗人员或者患者本人。
步骤S402：接收基于所述患者信息反馈的更正随访数据，并用所述更正随访数据替代所述数据异常对应的原始数据，以生成更正后的当前随访数据。
易于理解的是，在医疗人员或者患者本人通过用户端发送更正随访数据后，将错误的原始数据删除，用所述更正随访数据替代所述原始数据，生成更正的当前随访数据。
步骤S403：将更正后的当前随访数据输入训练后的长短期记忆网络注意力模型。
易于理解的是，将所述更正后的当前随访数据输入到训练后的长短期记忆网络注意力模型中，基于第一实施例，执行病情分析的功能，并将当前随访数据和病情分析结果作为所述随访记录收集结果。
本实施例通过训练后的长短期记忆网络注意力模型排查随访数据中的错误数据，并及时通知相关人员对错误数据进行更正，有助于降低其误诊率和漏诊率，根据更正的数据进行病情预测分析，有助于提升病情预测分析的准确率，辅助医疗人员获得更全面的随访数据。
参考图4，图4为本申请慢病随访记录收集方法第三实施例的流程示意图。
基于上述各实施例，在本实施例中，所述步骤S40中的在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析的步骤，可细化为：
步骤S404：在所述异常参数为病情异常参数时，获取所述病情异常参数对应的患者的历史随访数据。
易于理解的是，通过当前的随访数据获得了病情异常参数，进一步地需要结合历史随访数据对患者的病情进行程度判断。
步骤S405：根据预设病情变化信息及所述历史随访数据获取患者的病情变化程度。
需要说明的是，所述病情变化程度即患者病情属于病情中的何种阶段。所述预设病情变化信息程度为根据慢病数据获取的目标慢病的常见病变阶段，根据所述历史随访数据获取患者当前的病症状态，根据所述预设病情变化信息判断所述病症状态处于何种病情变化程度。例如：以脂肪肝的常见病变程度为例，可以分为轻度脂肪肝、中度及重度。获取到的历史随访数据为患者B超显示肝部由无脂肪堆积变化为有大量的脂肪堆积，所述病症状态为大量脂肪堆积，可以判断患者当前的病情变化程度为重度脂肪肝。
步骤S406：根据所述当前随访数据与所述病情变化程度进行病情预测，并将所述病情预测结果作为病情分析结果。
易于理解的是，根据所述当前随访数据与当前病情阶段结合训练后的长短期记忆网络注意力模型的输出结果中该患者特征对应的概率，可以对病情发展进行一定的预测。将所述病情预测结果作为病情分析结果。
本实施例通过对长短期记忆网络注意力模型的输出结果进行分析，不仅可以根据患者每次随访数据对患者病情进行评估和预测，还可以找出对病情影响较大的危险因素，便于患者及时的掌握自身身体状况，并对危险因素做出改善，更有利于个性化地预防病情恶化和并发症的发生。
此外，本申请实施例还提出一种存储介质，存储介质可以是易失性的，也可以是非易失性的，所述存储介质上存储有慢病随访记录收集程序，所述慢病随访记录收集程序被处理器执行时实现如下步骤：
根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
本申请存储介质的其他实施例或具体实现方式可参照上述各方法实施例，此处不再赘述。
参照图5，图5为本申请慢病随访记录收集装置第一实施例的结构框图。
如图5所示，本申请实施例提出的慢病随访记录收集装置包括：
集合生成模块501，用于根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
异常获取模块502，用于获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
模型训练模块503，用于获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
记录收集模块504，用于根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
本实施例通过历史随访数据对长短期记忆网络注意力模型进行训练，使得长短期记忆网络注意力模型能够符合收集随访数据的要求，使用训练后的长短期记忆网络注意力模型进行随访数据收集，不仅可以根据患者每次随访数据对患者病情进行评估和预测，还可以找出对病情影响较大的危险因素，便于患者及时的掌握自身身体状况，并对危险因素做出改善，更有利于个性化地预防病情恶化和并发症的发生。便于工作人员及时的向患者确认信息，提高数据库的质量。
基于本申请上述慢病随访记录收集装置第一实施例，提出本申请慢病随访记录收集装置的第二实施例。
集合生成模块501，还用于获取慢病随访过程中收集的患者慢病数据，根据所述患者慢病数据生成慢病随访表及慢病基线表；根据所述慢病随访表与所述慢病基线表生成慢病内容集合；根据慢病临床需求信息从所述慢病内容集合中获取慢病患者特征集合。
进一步地，异常获取模块502，还用于根据所述慢病基线表及所述慢病随访表确定所述慢病患者特征集合中各患者特征之间的特征关系；将所述特征关系输入预设规则引擎以获取特征逻辑错误条件；根据所述慢病基线表与所述慢病随访表确定特征预设范围条件；将所述特征逻辑错误条件、所述特征预设范围条件作为特征异常条件。
进一步地，模型训练模块503，还用于根据随访记录表获取预设时间内的历史随访数据；根据所述特征异常条件构建长短期记忆网络注意力模型；根据所述历史随访数据对应的时间点将所述历史随访数据输入所述长短期记忆网络注意力模型中进行模型训练。
进一步地，记录收集模块504，还用于根据所述随访记录表实时获取当前随访数据，并将所述当前随访数据输入训练后的长短期记忆网络注意力模型；根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数；在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析；将病情分析结果与所述当前随访数据作为随访记录收集结果。
进一步地，记录收集模块504，还用于在所述异常参数为数据异常参数时，获取所述数据异常参数对应的患者信息；接收基于所述患者信息反馈的更正随访数据，并用所述更正随访数据替代所述数据异常对应的原始数据，以生成更正后的当前随访数据；将更正后的当前随访数据输入训练后的长短期记忆网络注意力模型。
进一步地，记录收集模块504，还用于在所述异常参数为病情异常参数时，获取所述病情异常参数对应的患者的历史随访数据；根据预设病情变化信息及所述历史随访数据获取患者的病情变化程度；根据所述当前随访数据与所述病情变化程度进行病情预测，并将所述病情预测结果作为病情分析结果。
本申请慢病随访记录收集装置的其他实施例或具体实现方式可参照上述各方法实施例，此处不再赘述。
在另一个实施例中，本申请所提供的慢病随访记录收集方法，为进一步保证上述所有出现的数据的私密和安全性，上述所有数据还可以存储于一区块链的节点中。例如慢病患者特征集合及随访记录收集结果等，这些数据均可存储在区块链节点中。
需要说明的是，本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain)，本质上是一个去中心化的数据库，是一串使用密码学方法相关联产生的数据块，每一个数据块中包含了一批次网络交易的信息，用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者系统不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者系统所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者系统中还存在另外的相同要素。
上述本申请实施例序号仅仅为了描述，不代表实施例的优劣。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现，当然也可以通过硬件，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质（如只读存储器/随机存取存储器、磁碟、光盘）中，包括若干指令用以使得一台终端设备(可以是手机，计算机，服务器，空调器，或者网络设备等)执行本申请各个实施例所述的慢病随访记录收集方法。
以上仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种慢病随访记录收集方法，其中，所述慢病随访记录收集方法包括：
    根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
    获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
    获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
    根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
  2. 如权利要求1所述的慢病随访记录收集方法，其中，所述根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合的步骤，包括：
    获取慢病随访过程中收集的患者慢病数据，根据所述患者慢病数据生成慢病随访表及慢病基线表；
    根据所述慢病随访表与所述慢病基线表生成慢病内容集合；
    根据慢病临床需求信息从所述慢病内容集合中获取慢病患者特征集合。
  3. 如权利要求2所述的慢病随访记录收集方法，其中，所述获取所述慢病患者特征集合中不同患者特征对应的特征异常条件的步骤，包括：
    根据所述慢病基线表及所述慢病随访表确定所述慢病患者特征集合中各患者特征之间的特征关系；
    将所述特征关系输入预设规则引擎以获取特征逻辑错误条件；
    根据所述慢病基线表与所述慢病随访表确定特征预设范围条件；
    将所述特征逻辑错误条件、所述特征预设范围条件作为特征异常条件。
  4. 如权利要求3所述的慢病随访记录收集方法，其中，所述获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型的步骤，具体包括：
    根据随访记录表获取预设时间内的历史随访数据；
    根据所述特征异常条件构建长短期记忆网络注意力模型；
    根据所述历史随访数据对应的时间点将所述历史随访数据输入所述长短期记忆网络注意力模型中进行模型训练。
  5. 如权利要求4所述的慢病随访记录收集方法，其中，所述根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果的步骤，具体包括：
    根据所述随访记录表实时获取当前随访数据，并将所述当前随访数据输入训练后的长短期记忆网络注意力模型；
    根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数；
    在存在所述异常参数时，判断所述异常参数是否为病情异常参数；
    在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析；
    将病情分析结果与所述当前随访数据作为随访记录收集结果。
  6. 如权利要求5所述的慢病随访记录收集方法，其中，所述根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数的步骤之后，还包括：
    在所述异常参数为数据异常参数时，获取所述数据异常参数对应的患者信息；
    接收基于所述患者信息反馈的更正随访数据，并用所述更正随访数据替代所述数据异常对应的原始数据，以生成更正后的当前随访数据；
    将更正后的当前随访数据输入训练后的长短期记忆网络注意力模型。
  7. 如权利要求5所述的慢病随访记录收集方法，其中，所述在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析的步骤，具体包括：
    在所述异常参数为病情异常参数时，获取所述病情异常参数对应的患者的历史随访数据；
    根据预设病情变化信息及所述历史随访数据获取患者的病情变化程度；
    根据所述当前随访数据与所述病情变化程度进行病情预测，并将所述病情预测结果作为病情分析结果。
  8. 一种慢病随访记录收集装置，其中，所述慢病随访记录收集装置包括：
    集合生成模块，用于根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
    异常获取模块，用于获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
    模型训练模块，用于获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
    记录收集模块，用于根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
  9. 一种慢病随访记录收集设备，其中，所述设备包括：存储器、处理器及存储在所述存储器上并可在所述处理器上运行的慢病随访记录收集程序，所述慢病随访记录收集程序被处理器执行时实现如下步骤：
    根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
    获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
    获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
    根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
  10. 如权利要求9所述的慢病随访记录收集设备，其中，所述根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合的步骤，包括：
    获取慢病随访过程中收集的患者慢病数据，根据所述患者慢病数据生成慢病随访表及慢病基线表；
    根据所述慢病随访表与所述慢病基线表生成慢病内容集合；
    根据慢病临床需求信息从所述慢病内容集合中获取慢病患者特征集合。
  11. 如权利要求10所述的慢病随访记录收集设备，其中，所述获取所述慢病患者特征集合中不同患者特征对应的特征异常条件的步骤，包括：
    根据所述慢病基线表及所述慢病随访表确定所述慢病患者特征集合中各患者特征之间的特征关系；
    将所述特征关系输入预设规则引擎以获取特征逻辑错误条件；
    根据所述慢病基线表与所述慢病随访表确定特征预设范围条件；
    将所述特征逻辑错误条件、所述特征预设范围条件作为特征异常条件。
  12. 如权利要求11所述的慢病随访记录收集设备，其中，所述获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型的步骤，具体包括：
    根据随访记录表获取预设时间内的历史随访数据；
    根据所述特征异常条件构建长短期记忆网络注意力模型；
    根据所述历史随访数据对应的时间点将所述历史随访数据输入所述长短期记忆网络注意力模型中进行模型训练。
  13. 如权利要求12所述的慢病随访记录收集设备，其中，所述根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果的步骤，具体包括：
    根据所述随访记录表实时获取当前随访数据，并将所述当前随访数据输入训练后的长短期记忆网络注意力模型；
    根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数；
    在存在所述异常参数时，判断所述异常参数是否为病情异常参数；
    在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析；
    将病情分析结果与所述当前随访数据作为随访记录收集结果。
  14. 如权利要求13所述的慢病随访记录收集设备，其中，所述根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数的步骤之后，所述慢病随访记录收集程序被处理器执行时还实现如下步骤：
    在所述异常参数为数据异常参数时，获取所述数据异常参数对应的患者信息；
    接收基于所述患者信息反馈的更正随访数据，并用所述更正随访数据替代所述数据异常对应的原始数据，以生成更正后的当前随访数据；
    将更正后的当前随访数据输入训练后的长短期记忆网络注意力模型。
  15. 如权利要求13所述的慢病随访记录收集设备，其中，所述在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析的步骤，具体包括：
    在所述异常参数为病情异常参数时，获取所述病情异常参数对应的患者的历史随访数据；
    根据预设病情变化信息及所述历史随访数据获取患者的病情变化程度；
    根据所述当前随访数据与所述病情变化程度进行病情预测，并将所述病情预测结果作为病情分析结果。
  16. 一种存储介质，其中，所述存储介质上存储有慢病随访记录收集程序，所述慢病随访记录收集程序被处理器执行时实现如下步骤：
    根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合；
    获取所述慢病患者特征集合中不同患者特征对应的特征异常条件；
    获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型；
    根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果。
  17. 如权利要求16所述的存储介质，其中，所述根据慢病数据生成慢病内容集合，并根据所述慢病内容集合获取慢病患者特征集合的步骤，包括：
    获取慢病随访过程中收集的患者慢病数据，根据所述患者慢病数据生成慢病随访表及慢病基线表；
    根据所述慢病随访表与所述慢病基线表生成慢病内容集合；
    根据慢病临床需求信息从所述慢病内容集合中获取慢病患者特征集合。
  18. 如权利要求17所述的存储介质，其中，所述获取所述慢病患者特征集合中不同患者特征对应的特征异常条件的步骤，包括：
    根据所述慢病基线表及所述慢病随访表确定所述慢病患者特征集合中各患者特征之间的特征关系；
    将所述特征关系输入预设规则引擎以获取特征逻辑错误条件；
    根据所述慢病基线表与所述慢病随访表确定特征预设范围条件；
    将所述特征逻辑错误条件、所述特征预设范围条件作为特征异常条件。
  19. 如权利要求18所述的存储介质，其中，所述获取预设时间内的历史随访数据，根据所述历史随访数据与所述特征异常条件训练长短期记忆网络注意力模型的步骤，具体包括：
    根据随访记录表获取预设时间内的历史随访数据；
    根据所述特征异常条件构建长短期记忆网络注意力模型；
    根据所述历史随访数据对应的时间点将所述历史随访数据输入所述长短期记忆网络注意力模型中进行模型训练。
  20. 如权利要求19所述的存储介质，其中，所述根据训练后的长短期记忆网络注意力模型进行随访记录收集，获取随访记录收集结果的步骤，具体包括：
    根据所述随访记录表实时获取当前随访数据，并将所述当前随访数据输入训练后的长短期记忆网络注意力模型；
    根据训练后的所述长短期记忆网络注意力模型的输出结果检测所述当前随访数据是否存在异常参数；
    在存在所述异常参数时，判断所述异常参数是否为病情异常参数；
    在所述异常参数为病情异常参数时，获取所述病情异常对应的患者的历史随访数据，根据所述当前随访数据与所述历史随访数据对所述患者进行病情分析；
    将病情分析结果与所述当前随访数据作为随访记录收集结果。
PCT/CN2020/135272 2020-10-09 2020-12-10 慢病随访记录收集方法、装置、设备及存储介质 WO2021189958A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011074729.5A CN112201360B (zh) 2020-10-09 2020-10-09 慢病随访记录收集方法、装置、设备及存储介质
CN202011074729.5 2020-10-09

Publications (1)

Publication Number Publication Date
WO2021189958A1 true WO2021189958A1 (zh) 2021-09-30

Family

ID=74013200

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/135272 WO2021189958A1 (zh) 2020-10-09 2020-12-10 慢病随访记录收集方法、装置、设备及存储介质

Country Status (2)

Country Link
CN (1) CN112201360B (zh)
WO (1) WO2021189958A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115798714A (zh) * 2023-02-08 2023-03-14 南方医科大学南方医院 一种基于互联网的老年慢性病医疗判断方法及管理系统
US11723109B2 (en) 2020-10-21 2023-08-08 Ofinno, Llc Downlink data of small data transmission procedure

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113128359B (zh) * 2021-03-30 2021-11-16 广州华藏民族医药生物科技开发有限公司 一种用于检测草本濞舒膏疗效的图像特征识别系统
CN113643813B (zh) * 2021-08-30 2024-07-09 平安医疗健康管理股份有限公司 基于人工智能的慢病随访监管方法、装置及计算机设备
CN116779190B (zh) * 2023-06-25 2024-02-13 急尼优医药科技(上海)有限公司 一种基于物联网的医疗平台用户随访管理系统及方法

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107315906A (zh) * 2017-06-01 2017-11-03 北京瑞启医药信息科技有限公司 基于聊天机器人实现慢性病患者自动随访的方法及系统
CN107506598A (zh) * 2017-08-31 2017-12-22 深圳市易特科信息技术有限公司 基于健康检查一体机的慢病随访监控系统及方法
CN111524570A (zh) * 2020-05-06 2020-08-11 万达信息股份有限公司 一种基于机器学习的超声随访患者筛选方法

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190198174A1 (en) * 2017-12-22 2019-06-27 International Business Machines Corporation Patient assistant for chronic diseases and co-morbidities
US20190287685A1 (en) * 2018-03-16 2019-09-19 Vvc Holding Corporation String classification apparatus and methods using artificial intelligence

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107315906A (zh) * 2017-06-01 2017-11-03 北京瑞启医药信息科技有限公司 基于聊天机器人实现慢性病患者自动随访的方法及系统
CN107506598A (zh) * 2017-08-31 2017-12-22 深圳市易特科信息技术有限公司 基于健康检查一体机的慢病随访监控系统及方法
CN111524570A (zh) * 2020-05-06 2020-08-11 万达信息股份有限公司 一种基于机器学习的超声随访患者筛选方法

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ANONYMOUS: "Predictive Model Machine Learning Machine Learning | Survival Analysis of Follow-up Data", 7 February 2018 (2018-02-07), pages 1 - 11, XP055853379, Retrieved from the Internet <URL:https://www.sohu.com/a/221565935_743978> *
FAN ZHONG-HAO: "Early prediction of alzheimer's disease dementia based on baseline hippocampal MRI and 1-year follow-up cognitive measures using deep recurrent neural networks", 10 September 2019 (2019-09-10), pages 1 - 6, XP055853377, Retrieved from the Internet <URL:https://cloud.tencent.com/developer/article/1502757> *
WANG FEI , HUANG XIAO-HAN , WANG HONG QIAN: "Research on the Design and Integration of the Intelligent Follow-up Platform of Big Data of COVID-19", CHINA DIGITAL MEDICINE, vol. 15, no. 5, 15 May 2020 (2020-05-15), pages 73 - 75, XP055853384, ISSN: 1673-7571, DOI: 10.3969/j.issn.1673-7571.2020.05.025 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11723109B2 (en) 2020-10-21 2023-08-08 Ofinno, Llc Downlink data of small data transmission procedure
CN115798714A (zh) * 2023-02-08 2023-03-14 南方医科大学南方医院 一种基于互联网的老年慢性病医疗判断方法及管理系统
CN115798714B (zh) * 2023-02-08 2023-05-16 南方医科大学南方医院 一种基于互联网的老年慢性病医疗判断方法及管理系统

Also Published As

Publication number Publication date
CN112201360B (zh) 2023-06-20
CN112201360A (zh) 2021-01-08

Similar Documents

Publication Publication Date Title
WO2021189958A1 (zh) 慢病随访记录收集方法、装置、设备及存储介质
US8744867B2 (en) Method for selecting a clinical treatment plan tailored to patient defined health goals
US20050137910A1 (en) Systems and methods for automated extraction and processing of billing information in patient records
CN1745390A (zh) 自动提取及处理患者病历中记帐信息的系统和方法
US20100114599A1 (en) System for evaluation patient care outcomes
Lin et al. External validation of an algorithm to identify patients with high data-completeness in electronic health records for comparative effectiveness research
US7805421B2 (en) Method and system for reducing a data set
Dziadkowiec et al. Improving the quality and design of retrospective clinical outcome studies that utilize electronic health records
KR101565331B1 (ko) 패턴을 이용한 의료정보 분석시스템 및 방법
US20110153344A1 (en) Methods and apparatus for integrated medical case research and collaboration
Wu et al. Using standardised patients to assess the quality of medical records: an application and evidence from rural China
Lampe et al. The relationship of continuity of care, polypharmacy and medication appropriateness: a systematic review of observational studies
US20130282405A1 (en) Method for stepwise review of patient care
JP2020106882A (ja) 保険設計支援システム及び保険設計支援方法
US20170024784A1 (en) Computer System and Cost Calculating Method
Zamora et al. Characterizing chronic disease and polymedication prescription patterns from electronic health records
Tabisula Association rules in heart failure readmission rates and patient experience scores
JP2023020667A (ja) 医療情報処理方法、医療情報処理装置およびプログラム
Stakic et al. Secondary use of EHR data for correlated comorbidity prevalence estimate
Murata et al. A novel method for handling pre-existing conditions in prediction models for COVID-19 death
JP6094140B2 (ja) 診断プログラム、診断履歴作成方法、及び電子カルテ
Eftekhari et al. Impact of health information exchange adoption on physician’s referral behavior
JP7422651B2 (ja) 情報処理システム及び選択支援方法
US20230153757A1 (en) System and Method for Rapid Informatics-Based Prognosis and Treatment Development
JP7495554B1 (ja) 保険の引受査定を支援するための情報処理装置、情報処理方法及びプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20927000

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20927000

Country of ref document: EP

Kind code of ref document: A1