WO2023240837A1

WO2023240837A1 - Service package generation method, apparatus and device based on patient data, and storage medium

Info

Publication number: WO2023240837A1
Application number: PCT/CN2022/121728
Authority: WO
Inventors: 叶结保
Original assignee: 平安科技（深圳）有限公司
Priority date: 2022-06-15
Filing date: 2022-09-27
Publication date: 2023-12-21
Also published as: CN115171830A

Abstract

The present application relates to the technical field of big data. Disclosed are a service package generation method, apparatus and device based on patient data, and a storage medium. The method comprises: performing de-identification processing on collected original case data, so as to obtain target treatment data of a patient; extracting a plurality of key events from the target treatment data, and performing fusion processing on the key events, so as to obtain a medical information set of a disease corresponding to a disease type; inputting the target treatment data into a preset Bilstm model, so as to obtain a medical feature vector of medical data, and performing pooling analysis on the medical feature vector, so as to obtain a medical feedforward vector; and according to the medical feedforward vector, performing clustering analysis on the medical information set on the basis of a preset cosine similarity algorithm, so as to obtain a medical service package corresponding to the disease. By means of the present application, original case data is clustered to obtain service packages corresponding to different types of diseases with common features, thereby solving the technical problem of low medical service efficiency, and relieving the medical pressure.

Description

Service package generation method, device, equipment and storage medium based on patient data

This application requires the priority of the Chinese patent application submitted to the China Patent Office on June 15, 2022, with the application number 202210671458.4, and the invention name is "Service package generation method, device, equipment and storage medium based on patient data", which The entire contents are incorporated herein by reference.

Technical field

This application relates to the field of big data technology, and in particular to a method, device, equipment and storage medium for generating a service package based on patient data.

Background technique

The current medical situation in our country is basically disease-specific treatment, that is, users find themselves sick and go to the hospital to see a doctor. After the doctor makes a diagnosis, he or she will prescribe medication, review and other treatment plans for the patient, and the patient will cooperate with the treatment according to the treatment plan. Once the patient recovers, the visit is considered over. The inventor realized that this method has the following shortcomings:

1. The traditional medical treatment model does not allow patients to participate in their own health management, understand the causes of disease, and popularize knowledge about disease prevention, and there is no prevention and treatment services for disease-related complications.

2. The traditional medical treatment model is a passive medical treatment model. There is no mechanism for early prevention. There is no improvement in patient health.

3. The traditional medical treatment model is basically that the doctor treats a single disease without complications, and provides comprehensive prevention and treatment of potential diseases.

At present, only a small number of rich people have their own dedicated family doctors, and most ordinary people do not have comprehensive health management capabilities. How to allow most people to have early prevention when they are not sick, receive comprehensive treatment when they are sick, and get comprehensive management of health is a common social problem. Therefore, how to provide patients with convenient medical treatment based on patient data and improve the efficiency of patient treatment has become a technical problem that needs to be solved by those skilled in the art.

Contents of the invention

The main purpose of this application is to solve the technical problem in the existing technology of being unable to analyze original case data and obtain service packages corresponding to different types of diseases, and to improve the efficiency of medical services.

The first aspect of this application provides a method for generating service packages based on patient data, including: collecting original case data of similar diseases from a preset medical information platform, and extracting disease types and treatment data in the original case data; The treatment data is de-identified to obtain target treatment data; multiple key events in the target treatment data are extracted, and the key events are fused to obtain medical information of the disease corresponding to the disease type. Set; input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector; according to the medical feedforward vector Feed vector, perform cluster analysis on the medical information collection based on a preset cosine similarity algorithm, and obtain a medical service package corresponding to the disease.

A second aspect of this application provides a device for generating a service package based on patient data, including a memory and at least one processor, instructions are stored in the memory, and the memory and the at least one processor are interconnected through lines; The at least one processor calls the instructions in the memory. When the processor executes the computer-readable instructions, the following steps are implemented: collect original case data of similar diseases from the preset medical information platform, and extract all the original case data. Describe the disease type and treatment data in the original case data; de-identify the treatment data to obtain target treatment data; extract multiple key events in the target treatment data, and perform fusion processing on the key events , obtain the medical information set of the disease corresponding to the disease type; input the medical information set into the preset Bilstm model for vector calculation, obtain the medical feature vector of the disease, and pool the medical feature vector Analyze to obtain a medical feedforward vector; perform cluster analysis on the medical information set based on the preset cosine similarity algorithm according to the medical feedforward vector to obtain a medical service package corresponding to the disease.

A third aspect of the present application provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is run on a computer, it causes the computer to perform the following steps: from preset medical information to The platform collects original case data of similar diseases and extracts disease types and treatment data in the original case data; de-identifies the treatment data to obtain target treatment data; extracts multiple target treatment data in the target treatment data. key events, and the key events are fused to obtain the medical information set of the disease corresponding to the disease type; the medical information set is input into the preset Bilstm model for vector calculation, and the medical information set of the disease is obtained. feature vectors, and pooling analysis is performed on the medical feature vectors to obtain medical feedforward vectors; according to the medical feedforward vectors, clustering analysis is performed on the medical information collection based on a preset cosine similarity algorithm to obtain the Medical service package corresponding to the above-mentioned diseases.

The fourth aspect of this application provides a device for generating service packages based on patient data, including: a collection module for collecting original case data of similar diseases from a preset medical information platform, and extracting the original case data from the original case data. Disease type and treatment data; a de-identification module, used to de-identify the treatment data to obtain target treatment data; a fusion module, used to extract multiple key events in the target treatment data, and combine the The key events are fused to obtain the medical information set of the disease corresponding to the disease type; the pooling module is used to input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease. and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector; a clustering module is used to perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm according to the medical feedforward vector. , get the medical service package corresponding to the disease.

In the technical solution provided by this application, original case data of similar diseases are collected from a preset medical information platform, and the disease types and treatment data in the original case data are extracted; and the treatment data are de-identified. Obtain target treatment data; extract multiple key events in the target treatment data, and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type; input the target treatment data into a pre-set Assume that in the Bilstm model, the medical feature vector of the medical data is obtained, and the medical feature vector is pooled and analyzed to obtain a medical feedforward vector; according to the medical feedforward vector, all the medical feature vectors are calculated based on the preset cosine similarity algorithm. Perform cluster analysis on the medical information collection to obtain a medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

Description of the drawings

Figure 1 is a schematic diagram of the first embodiment of the service package generation method based on patient data provided by this application;

Figure 2 is a schematic diagram of the second embodiment of the service package generation method based on patient data provided by this application;

Figure 3 is a schematic diagram of the third embodiment of the service package generation method based on patient data provided by this application;

Figure 4 is a schematic diagram of the fourth embodiment of the service package generation method based on patient data provided by this application;

Figure 5 is a schematic diagram of the fifth embodiment of the service package generation method based on patient data provided by this application;

Figure 6 is a schematic diagram of a first embodiment of a service package generation device based on patient data provided by this application;

Figure 7 is a schematic diagram of a second embodiment of a service package generation device based on patient data provided by this application;

Figure 8 is a schematic diagram of an embodiment of a service package generation device based on patient data provided by this application.

Detailed ways

The method, device, equipment and storage medium for generating service packages based on patient data provided by the embodiments of this application first collect original case data of similar diseases from a preset medical information platform, and extract the diseases in the original case data. type and treatment data; de-identify the treatment data to obtain target treatment data; extract multiple key events in the target treatment data, and fuse the key events to obtain the disease type A collection of medical information corresponding to the disease; input the target treatment data into the preset Bilstm model to obtain the medical feature vector of the medical data, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector; according to The medical feedforward vector performs cluster analysis on the medical information set based on a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

The terms "first", "second", "third", "fourth", etc. (if present) in the description and claims of this application and the above-mentioned appendix are used to distinguish similar objects and are not necessarily used to describe A specific order or sequence. It is to be understood that the data so used are interchangeable under appropriate circumstances so that the embodiments described herein can be practiced in sequences other than those illustrated or described herein. In addition, the terms "comprising" or "having" and any variations thereof are intended to cover non-exclusive inclusions, e.g., processes, methods, systems, products, or devices that comprise a series of steps or units and are not necessarily limited to those expressly listed. steps or units, but may include other steps or units not expressly listed or inherent to such processes, methods, products or apparatuses.

For ease of understanding, the specific process of the embodiment of the present application is described below. Please refer to Figure 1. The first embodiment of the method for generating a service package based on patient data in the embodiment of the present application includes:

101. Collect original case data of similar diseases from the preset medical information platform, and extract disease types and treatment data from the original case data;

In this embodiment, when patients go to the Internet hospital built by Ping An Health to see a doctor, a large amount of data will be generated, including patient big data, diagnosis and treatment big data, disease big data, etc.

When designing the business system, the product will add the data features required by the business to the interface:

For example, when a patient consults, he must first fill in the diagnosis and treatment card and medical record information and then conduct the consultation. The doctor will prescribe or provide care according to the patient's condition. Each item will fill in relevant information on the page. This information is the characteristic information that can be summarized by the physical examination:

a. Diagnosis and treatment card information: patient name, gender, age, birthday, household registration address, residential address, ID type, ID number, mobile phone number, social security card number, ethnicity, nationality, guardian information (name, phone number, ID card, etc.), marriage status, etc.

b. Patient condition information: affected area, related living habits, condition description, incidence pattern, allergy history, medication history, illness time, condition grade, etc.

c. Consultation and diagnosis information: name of disease, name of suspected disease, severity of illness, consultation suggestions, etc.

d. Doctor’s prescribing information: department, diagnosis results, prescription category, drug name, drug specifications, drug brand, drug usage

e. Nursing information: disease name, condition information, nursing items (moxibustion, massage, physiotherapy, etc.), number of nursing sessions, etc.

102. De-identify the treatment data to obtain target treatment data;

In this embodiment, de-identification refers to a data processing method that processes identifiers so that the processed information cannot identify a specific personal information subject. Among them, the biggest difference between China's "Personal Information Security Specifications" and "Personal Information De-Identification Guidelines" and the definition of de-identification in relevant laws in the United States, Canada and other regions is whether the possibility of indirect identification needs to be considered to prevent re-identification. China limits re-identification to "without the use of additional information", which denies the situation of "indirect identification", which is very similar to pseudonymization in GDPR; laws such as CCPA and HIPAA have higher requirements for preventing re-identification. The possibility of re-identification needs to be comprehensively evaluated in conjunction with other additional information that may be available.

Specifically, de-identification refers to the process of removing the association between a set of identifiable data and the data subject. Through this process, data managers can delete or change the identification information in the data set, making it difficult or impossible for attackers to use the data set to identify specific individual subjects, so that the data set can be shared for use within a predetermined range. De-identification is one of the main tools of privacy preserving data publishing (PPDP) [1]. By removing the association between privacy attributes and data subjects in the data set, and having sufficient ability to prevent re-identification, Certain attributes of the data set can be shared and published for processing and analysis by external business systems.

103. Extract multiple key events in the target treatment data, and fuse the key events to obtain a collection of medical information for the disease corresponding to the disease type;

In this embodiment, key events may refer to core events that are set according to different diseases of the target object in the treatment data and can represent the diagnosis and treatment record or the corresponding disease in the diagnosis and treatment plan. For example, it is assumed that the disease of the target object in the treatment data can be Breast cancer is a breast cancer disease. Since the treatment strategies for breast cancer diseases at different clinical stages are completely different, the treatment options for breast cancer diseases at the same clinical stage will also be different based on different molecular classifications and pathological diagnosis. For example, targeting HER2 (human epidermal growth factor) For breast cancer patients who are positive for breast cancer (factor receptor), the timing of administration of the targeted drug trastuzumab and the judgment of its therapeutic efficacy are particularly critical. Therefore, it is necessary to set key events of breast cancer disease to facilitate the completion of scientific research and facilitate retrospective analysis. Survival analysis of different treatment options for breast cancer patients to further optimize treatment strategies for breast cancer patients; specifically, combined with the characteristics of breast cancer disease, the key events in the treatment data can be the first diagnosis and treatment event, or local treatment-related events, Of course, it can also be drug treatment-related events, efficacy evaluation events, adverse drug reaction events, etc., and this example embodiment is not limited to this.

In this embodiment, data fusion technology refers to information processing technology that uses computers to automatically analyze and synthesize certain observation information obtained in time series under certain criteria to complete the required decision-making and evaluation tasks. Data fusion technology includes the collection, transmission, synthesis, filtering, correlation and synthesis of useful information from various information sources to assist people in situation/environment determination, planning, detection, verification, and diagnosis.

104. Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain the medical feedforward vector;

In this embodiment, the target treatment data is converted from text form to vector form for final calculation of the similarity between vectors, and the medical word that best matches the sentence to be analyzed is determined from multiple medical words based on the similarity; The Bilstm model is a natural language processing neural network model. The Bilstm model converts target treatment data into vectors. For example, the sentence to be analyzed is "eye swelling, eye pain, photophobia, hard eyeball, weak vision"; the first medical word is "glaucoma, acute angle-closure glaucoma, chronic angle-closure glaucoma, primary open-angle glaucoma, filter "Overbubble separation", the second medical word "myopia", the third medical word "keratitis", etc.

Among them, the Bilstm model splices the target treatment data to obtain the original sentence vector HL, that is, the original sentence vector HL {eye swelling + eye pain + photophobia + hard eyeball + weak vision}, etc.

The pooling analysis is based on the pooling proposed by the convolutional neural network. In this embodiment, maxpooling and avg pooling are used to process the target treatment data respectively, and a combination of maxpooling and maxpooling are used. ) and mean pooling (avg pooling) perform pooling operations in parallel dual pooling layers to retain deeper semantic information of the target treatment data.

105. According to the medical feedforward vector, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm to obtain services corresponding to the disease.

In this embodiment, cosine similarity, also known as cosine similarity, evaluates the similarity between two vectors by calculating the cosine value of the angle between them. Cosine similarity draws vectors into vector space according to coordinate values, such as the most common two-dimensional space.

Cosine similarity measures the similarity between two vectors by measuring the cosine of their angle. The cosine of an angle of 0 degrees is 1, while the cosine of any other angle is no greater than 1; and its minimum value is -1. The cosine of the angle between two vectors thus determines whether the two vectors point roughly in the same direction. When two vectors point in the same direction, the cosine similarity value is 1; when the angle between the two vectors is 90°, the cosine similarity value is 0; when the two vectors point in completely opposite directions, the cosine similarity value is -1. This result has nothing to do with the length of the vector, only the direction in which the vector points. Cosine similarity is usually used in positive spaces and therefore gives values between -1 and 1.

Note that these upper and lower bounds apply to vector spaces of any dimension, and cosine similarity is most commonly used in high-dimensional positive spaces. For example, in information retrieval, each term is assigned a different dimension, and a dimension is represented by a vector, whose values in each dimension correspond to the frequency of occurrence of the term in the document. Cosine similarity thus gives how similar two documents are with respect to their subject matter. Additionally, it is often used for file comparison in text mining. In addition, in the field of data mining, it is used to measure the cohesion within the cluster.

Among them, cluster analysis refers to the analysis process of grouping a collection of physical or abstract objects into multiple classes composed of similar objects. The goal is to collect data to classify based on similarity. Clustering originates from many fields, including mathematics, computer science, statistics, biology, and economics. In different application fields, many clustering techniques have been developed. These technical methods are used to describe data, measure the similarity between different data sources, and classify data sources into different clusters.

In this embodiment, the service package is also called a card product. Specifically, based on the obtained target treatment data, basic services are first created, and these services are combined into service packages. Among them, a service package includes one or more services, such as disease course management, Sanfu moxibustion, COPD, ECG monitoring, children's eye examination, quick consultation, worry-free hospitalization, etc.

Multiple services in a service package are often a combination of one or a group of services that are closely related. For example, the Sanfu moxibustion service package includes 5 services of Sanfu moxibustion, which is specially designed for the elderly to treat winter diseases in summer and prevent them in advance. Another example: the cardiology department inpatient secretary service package is specifically for patients with cardiology diseases. It includes three services: cardiology department offline diagnosis green pass, cardiology department offline examination green pass, and cardiology inpatient green pass.

Service package classification: Each service package contains one or more classification labels. The classification labels are selected from various common phenomena, such as children, teenagers, youth, middle-aged, and elderly according to age groups. Classified by department: ophthalmology, cardiology, thoracic surgery, etc. The applicable groups are divided into: students, office workers, pregnant women, people with high income, etc.

In the embodiment of the present application, original case data of similar diseases are collected from a preset medical information platform, and the disease type and treatment data in the original case data are extracted; the treatment data is de-identified to obtain the target Treatment data; extract multiple key events in the target treatment data, and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type; input the target treatment data into the preset Bilstm In the model, the medical feature vector of the medical data is obtained, and the medical feature vector is pooled and analyzed to obtain a medical feedforward vector; according to the medical feedforward vector, the medical feature vector is calculated based on a preset cosine similarity algorithm. The information collection is subjected to cluster analysis to obtain the medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

Please refer to Figure 2. The second embodiment of the service package generation method based on patient data in the embodiment of this application includes:

201. Collect original case data of similar diseases from the preset medical information platform, and extract disease types and treatment data in the original case data;

202. Preset diagnosis and treatment templates for different disease types;

In this embodiment, before collecting original case data of similar diseases from the preset medical information platform and extracting disease types and treatment data in the original case data, different diagnosis and treatment templates are configured for different types of patients. Different types of patients specifically refer to patients with different reasons for seeking medical treatment. When patients seek medical treatment for different reasons (specific reasons include: chest pain, stroke, and trauma), they generally have corresponding diagnosis and treatment procedures for examination, diagnosis, and treatment. Therefore, a diagnosis and treatment template can be constructed based on the corresponding diagnosis and treatment procedures for subsequent entry. Chronological verification of information.

203. Obtain the time information corresponding to the treatment data;

In this embodiment, an information entry template is configured, and various treatment data of patients entered by the user are collected through the information entry template.

Before obtaining treatment data, configure an information entry template. Users can enter information according to the prompts of the pre-configured information entry template, collect patient information, reduce errors and omissions in information entry, and reduce the frequency of subsequent modifications. Specifically, the information entry template includes sub-templates for each diagnosis and treatment item for each type of patient.

Preferably, the time information of each treatment data is obtained, specifically: the input time of each treatment data entered by the patient as the corresponding time information is obtained. Generally speaking, when patients undergo various diagnosis and treatment projects, various treatment data will be entered simultaneously. Therefore, the entry time of the treatment data can be used as the time information corresponding to the treatment data to realize automatic entry of time information and reduce workload. . At the same time, after the time information is automatically entered, the user can also be prompted to check the automatically entered time information and modify it if it is wrong.

204. According to the disease type, obtain the diagnosis and treatment template corresponding to the treatment data;

In this embodiment, the diagnosis and treatment template is generated based on the patient's disease type and its sequence requirements. Specifically, this embodiment sets a diagnosis and treatment template corresponding to STEMI type patients. The diagnosis and treatment template includes onset, first ECG collection in the hospital, first ECG diagnosis in the hospital, starting informed consent, starting the catheterization laboratory, activating the catheterization laboratory, and the patient arriving at the catheter There are ten diagnosis and treatment items in the laboratory, deciding on interventional surgery, signing informed consent, and passing the guide wire, and stipulates three sequence requirements. The three sequence requirements are: the time to start informed consent should be later than the time to start the cath lab, and the time to start the cath lab. The time to decide on interventional surgery should be later than the time to sign informed consent, and the time to sign informed consent should be later than the time to start informed consent. Verify the treatment data of corresponding types of patients based on this diagnosis and treatment template.

205. Verify the treatment data according to the diagnosis and treatment template to determine whether the time sequence of the treatment data is correct;

In this embodiment, the diagnosis and treatment template also includes duration indicators for each diagnosis and treatment item; calculates the actual duration of each diagnosis and treatment item based on the time information of each treatment data of the current patient; and determines whether the actual duration meets the corresponding requirements. Duration indicator, if so, report various treatment data, otherwise a warning will be given.

In addition to verifying the time sequence requirements of each diagnosis and treatment item, this embodiment also verifies the duration requirements of each diagnosis and treatment item. When setting up a STEMI type patient diagnosis and treatment template, set the duration indicator for each diagnosis and treatment item, such as greater than the set time, less than the set time, or within the set time range. Then, based on the time information of the entered treatment data, the actual duration of each diagnosis and treatment item is calculated, and whether the actual duration meets the duration indicator is determined.

206. De-identify the treatment data to obtain target treatment data;

207. Extract multiple key events in the target treatment data, and fuse the key events to obtain a collection of medical information for the disease corresponding to the disease type;

208. Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain the medical feedforward vector;

209. According to the medical feedforward vector, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm, and obtain the medical service package corresponding to the disease.

Steps 201 and 206-209 in this embodiment are similar to steps 101 and 102-105 in the first embodiment, and will not be described again here.

In the embodiment of this application, original case data of similar diseases are collected from the preset medical information platform, and the disease type and treatment data in the original case data are extracted; the treatment data is de-identified to obtain the target treatment data; extraction Multiple key events in the target treatment data, and fuse the key events to obtain a medical information collection of diseases corresponding to the disease type; input the target treatment data into the preset Bilstm model to obtain the medical feature vector of the medical data, and Pooling analysis is performed on the medical feature vectors to obtain medical feedforward vectors; according to the medical feedforward vectors, cluster analysis is performed on the medical information collection based on the preset cosine similarity algorithm to obtain a medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

Please refer to Figure 3. The third embodiment of the service package generation method based on patient data in the embodiment of this application includes:

301. Collect original case data of similar diseases from the preset medical information platform, and extract disease types and treatment data from the original case data;

302. Construct a treatment data query database based on original case data;

In this embodiment, based on the disease impact characteristics and clinical treatment data, combined with the patient's treatment data from multiple databases of HIS, RIS, and PACS, a treatment data query database is constructed, and a differential privacy algorithm and an encryption algorithm are used to combine the patient's treatment data. The private data is de-identified to update the treatment data query database, so that a clinical image query and diagnosis and treatment system can be established based on the updated treatment data query database. The embodiment of this application not only meets the confidentiality requirements in the differential privacy protection model, but also ensures that the database The reliability of the data released in the system can help clinical researchers query and collect past cases, big data analysis and evaluation, and lay a good foundation for promoting the automation of medical data statistics, eliminating information islands, and providing decision support.

303. According to the differential privacy algorithm, add random noise to the sensitive attribute fields in the data table of the treatment data query database;

In this embodiment, the differential privacy algorithm adds an appropriate amount of noise to the statistical results to ensure that modifying an individual record in the data set will not have a significant impact on the statistical results, thus meeting the requirements for privacy protection.

Assume that D ₁ and D ₂ are adjacent data sets, S is all possible outputs of random function A, and Pr is the probability that A (D ₁ ) obtains a certain value. Then as long as the algorithm satisfies the following formula, it can be said that this algorithm satisfies ε -The standard for differential privacy.

Pr[A(D ₁ )∈S]≤e∈×Pr[A(D ₂ )∈S]

Among them, the probability Pr[·] represents the risk of privacy leakage, which is controlled by the randomness of algorithm A(D); ε is the privacy protection budget parameter, which is used to adjust and balance data privacy security and data reliability. By adding random noise To achieve privacy protection, that is, the smaller ε is, the greater the noise is added, and the higher the degree of privacy protection. Similarly, the larger ε is, the smaller the noise is added, and the weaker the privacy protection security is.

Optionally, random noise is added using different noise mechanisms to the raw treatment data for sensitivity fields of different data types. For example, the Laplacian mechanism is used to add random noise to the original treatment data of numeric type sensitivity fields, and the exponential mechanism is used to add random noise to the original treatment data of non-numeric type sensitivity fields.

The Laplacian mechanism processes numerical data (continuous data), such as patient age, and adds random noise to the numerical results to achieve differential privacy. The exponential mechanism processes non-numeric (discrete data) data, and does not return deterministic results, but returns results with a certain probability value. The output is a set of discrete data, which can be determined by the scoring function. The output with a high score has a high probability, and the score Low output probability is low.

Specifically, Laplacian noise is added to numerically sensitive attributes such as patient age and examination date in the table, and exponential noise is added to attributes such as gender, education level, region, examination equipment type, and disease in the data table to obtain the noise results. Replace it in the data table.

304. Based on random noise, de-identify the sensitive attribute fields in the original case data to obtain the identifier field;

In this embodiment, the original database structure is roughly divided into the following categories: display identifier: a set of attributes that can uniquely identify a single individual, such as the patient name and patient number fields in the table; sensitive attributes: an attribute set containing private data: patient gender , age, education level, region, type of inspection equipment, disease, etc.; non-sensitive attributes: attribute sets in addition to the above categories.

305. Encrypt the identifier field to obtain the patient's target treatment data;

In this embodiment, the identifier field of the data table is encrypted according to the data encryption standard DES encryption algorithm and Base64 encoding; DES symmetric encryption is a relatively traditional encryption method and has extremely high security. Base64 is a representation of binary data based on 64 printable characters. These two methods are combined to meet the encryption requirements for uniquely identified sensitive attributes such as patient names, examination numbers, and image numbers.

306. Extract multiple key events in the target treatment data, and fuse the key events to obtain a collection of medical information for the disease corresponding to the disease type;

307. Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain the medical feedforward vector;

308. According to the medical feedforward vector, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm, and obtain the medical service package corresponding to the disease.

Steps 301, 306-308 in this embodiment are similar to steps 101, 103-105 in the first embodiment, and will not be described again here.

In the embodiment of this application, original case data of similar diseases are collected from the preset medical information platform, and the disease type and treatment data in the original case data are extracted; the treatment data is de-identified to obtain the target treatment data; extraction Target multiple key events in the treatment data and fuse the key events to obtain a medical information set of diseases corresponding to the disease type; input the target treatment data into the preset Bilstm model to obtain the medical feature vector of the medical data , and perform pooling analysis on the medical feature vectors to obtain medical feedforward vectors; according to the medical feedforward vectors, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm, and obtain the medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

Please refer to Figure 4. The fourth embodiment of the service package generation method based on patient data in the embodiment of this application includes:

401. Collect original case data of similar diseases from the preset medical information platform, and extract disease types and treatment data in the original case data;

402. De-identify the treatment data to obtain target treatment data;

403. Obtain the predefined key event collection;

In this embodiment, a predefined key event set is obtained; multiple key events corresponding to the target object are extracted from the treatment data according to the key event set. The predefined key event set may refer to a set of key events corresponding to the disease set by the staff based on the special characteristics of various diseases in different clinical stages. That is, the predefined key event set may include the corresponding key events of the target disease. The key events of different clinical stages are extracted from the treatment data based on the predefined key event set to extract multiple key events corresponding to the target object. By predefining key event sets, the accuracy of multiple key events extracted can be effectively ensured, while the extraction efficiency of key events can be ensured.

Preferably, the predefined key event set can be updated periodically, or it can be updated immediately when new key events occur. Specifically, it can be updated manually or through crawler tools or artificial intelligence. This example implements There are no special restrictions on this.

404. Filter the target treatment data to filter invalid medical data in the target treatment data and obtain valid medical data;

In this embodiment, the treatment data is screened to filter invalid medical data in the treatment data; multiple key events corresponding to the target object are extracted from the filtered treatment data according to the key event set. Invalid medical data may refer to treatment data in the treatment data that is meaningless for the diagnosis and treatment of the target disease of the target object. For example, invalid medical data may be treatment data corresponding to non-target diseases of the target object in the treatment data, or it may refer to treatment data. Incomplete treatment data corresponding to the target disease in the treatment (for example, treatment data that ends halfway through treatment due to special reasons). Of course, it can also be other treatment data in the treatment data that are meaningless for the diagnosis and treatment of the target disease of the target object. This example embodiment does not impose special limitations on this. By filtering invalid medical data in treatment data, the accuracy of treatment data can be effectively improved, further ensuring the accuracy of extraction of key events, reducing unnecessary calculations, and saving computing resources.

405. Extract effective medical data according to the key event set to obtain multiple key events in the target treatment data;

In this embodiment, in an example embodiment, the second key event may refer to an event determined by extracting key events from multiple treatment data as a whole. For example, the second key event may be all treatment data of the corresponding series of the target object. The first diagnosis and treatment event in , can also be an adverse drug reaction event in all treatment data. Of course, the second key event can also be an event determined by joint judgment on treatment data recorded across multiple visits, such as for the first time after relapse. For chemotherapy events, first determine the first visit for the first recurrence, then extract chemotherapy drugs from the drug orders of all visits after this visit, and finally find the first visit where chemotherapy drugs appear, then the first chemotherapy event after recurrence can be determined. Of course, This is only a schematic illustration, and should not impose any special limitations on this exemplary embodiment.

In another example embodiment, the treatment data of the target subject are jointly constituted in chronological order according to the first key event extracted from the sorted single treatment data and the second key event extracted from the sorted plurality of treatment data. corresponding key events. By constructing key events from the first key event extracted from the sorted single treatment data and the second key event extracted from the sorted multiple treatment data, it is possible to avoid the need for insufficient screening or the need to span across The treatment data recorded in multiple visits are jointly judged to determine the events, which will lead to the problem of missing key events or inaccurate key events, thereby improving the accuracy of key events.

406. Determine the attribute characteristics corresponding to the key events and the weight values of the attribute characteristics;

In this embodiment, the attribute characteristics corresponding to the key events and the weight values of the attribute characteristics are determined. Among them, the weight value of the attribute feature is determined; the score data of multiple target key events are determined based on the attribute feature and the weight value, and the multiple target key events are deduplicated based on the score data to filter target key events with the same meaning. The weight value can refer to the weight of a preset attribute feature. For example, the weight of the time attribute can be 0.3, and the weight of the recurrence type can be 0.7. Of course, the weight value of the attribute feature can be customized according to the actual situation. This example implements There are no special restrictions on this. Score data can refer to similarity data used to determine whether two key events are similar key events. The higher the score data of two key events, the two key events are considered to be similar key events, and only one of them is retained. Can.

407. Determine the score data of key events according to the attribute characteristics and weight values, and obtain the medical information collection of the disease corresponding to the disease type based on the score data;

In this embodiment, fusion processing may refer to a process of constructing multiple key events into target data in a specific order. For example, fusion processing may include but is not limited to arranging multiple key events in chronological order. In this example embodiment, There are no special restrictions on this. Target data may refer to structured data generated by fusion processing of multiple key events and corresponding to the treatment data of the target object.

Before fusing multiple key events, first determine the attribute characteristics corresponding to the multiple key events. Attribute characteristics can refer to different attributes corresponding to key events. For example, attribute characteristics can be the time attribute corresponding to key events, or the recurrence type corresponding to key events. Of course, attribute characteristics can also be other attributes corresponding to key events. In this example The embodiment does not specifically limit this.

408. Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain the medical feedforward vector;

409. According to the medical feedforward vector, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm, and obtain the medical service package corresponding to the disease.

Steps 401-402 and 408-409 in this embodiment are similar to steps 101-102 and 104-105 in the first embodiment, and will not be described again here.

In the embodiment of this application, the original case data of similar diseases is collected from the preset medical information platform, and the disease type and treatment data in the original case data are extracted; the treatment data is de-identified to obtain the target treatment data; Extract multiple key events in the target treatment data and fuse the key events to obtain a medical information collection of diseases corresponding to the disease type; input the target treatment data into the preset Bilstm model to obtain the medical feature vector of the medical data. The medical feature vectors are pooled and analyzed to obtain medical feedforward vectors; according to the medical feedforward vectors, the medical information collection is clustered and analyzed based on the preset cosine similarity algorithm to obtain the medical service package corresponding to the disease. This application performs cluster analysis on original case data to obtain target treatment data with common characteristics, and generates service packages corresponding to different types of diseases based on the target treatment data, solving the problem of the lack of sustainable tracking services in current medical services and the difficulty in achieving timely The lack of effective feedback and effective integration of medical data information leads to the problem that the quality of mobile medical services cannot be guaranteed. Effectively improve the efficiency and portability of medical services, reduce waiting time for medical personnel, and relieve medical pressure.

Please refer to Figure 5. The fifth embodiment of the service package generation method based on patient data in the embodiment of this application includes:

501. Collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

502. De-identify the treatment data to obtain target treatment data;

503. Extract multiple key events in the target treatment data, and fuse the key events to obtain a medical information collection of diseases corresponding to the disease type;

504. Extract features from the medical feature vector to obtain the target medical features of the medical feature vector;

In this embodiment, the medical feature vector is input into two different parallel pooling layers for dimensionality reduction after passing through convolutional layers with different window sizes and feature extraction of filtering units below them - avg pooling layer and max pooling layer. , fully combines the dynamic extraction characteristics of the max pooling layer and the contribution ability of the avg pooling layer to the average semantics of short texts, effectively reducing the loss of semantic information during dimensionality reduction; finally, the necessary semantic splicing is performed in the concatenation layer to form Original feedforward vector and medical feedforward vector; among them, the max pooling layer dynamic extraction method and the avg pooling layer average semantic method for short text take into account the impact of the height of the convolution kernel sliding window on the generated feature map. That is, the height of the convolution kernel is used as an important basis for the number M of down-sampling of the feature map. The higher the convolution kernel, the fewer the number of down-sampling. On the contrary, the lower the height of the convolution kernel, the greater the number of down-sampling.

505. Obtain the frequency of target medical features;

In this embodiment, the frequency of the target medical feature may be the frequency of occurrence of the target medical feature in different historical analysis models. For different target medical features, their frequency of occurrence in multiple historical analysis models can be different. For example, the surgical fee may appear in a variety of different historical analysis models with higher frequency, while the glucose content may only appear in the diabetes surgery analysis model with lower frequency.

Specifically, the higher the frequency of the target medical feature appearing in different historical analysis models. If the target medical feature appears in all historical analysis models, it can be determined that the more important the target medical feature is. Thus, target medical features can be determined based on the frequency of features appearing in different historical analysis models.

Frequency compliance requirements can be that the frequency of the target medical features appearing in the historical analysis model meets certain threshold conditions, or the frequency of the target medical features appearing in the historical analysis model is sorted, and the target medical features that meet certain requirements are ranked as frequency compliance requirements. target medical characteristics.

506. Extract the first data corresponding to the target medical feature in the target treatment data;

In this embodiment, multidimensional data may refer to all data stored in the database, and may include data newly added every time the data is changed as well as historical data before the change. For example, corresponding to the medical insurance data mentioned above, the initial data refers to the user data. The medical data generated after medical treatment and stored under the user's name can include historical medical data and current medical data. Specifically, it can include but is not limited to consultation location, consultation time, International Classification of Diseases (ICD), registration Department, registered doctor information, registration fee, payment method, examination items, examination fee, condition description, medical treatment suggestions, drug list, drug price, drug dosage, payment window, drug collection window, whether to return for follow-up consultation, time for follow-up consultation, number of consultations, etc. data.

Specifically, the server can extract initial data from multi-dimensional data based on the selected target medical features. The extracted initial data can be divided into multiple categories. For example, for medical insurance data, it can include but is not limited to this medical expense data, this time Medical ICD data, historical medical data. Among them, the cost data of this medical treatment can include but is not limited to surgery fees, drug fees, examination fees, etc.; the ICD data of this medical treatment can include but is not limited to the cost of this confirmed ICD, the average cost of the ICD, etc.; historical medical data can Including but not limited to data such as the number of local outpatient clinics, the number of local hospitalizations, the number of outpatient clinics in other places, the number of hospitalizations in other places, the proportion of local outpatient visits, the proportion of outpatient visits in other places, etc.

507. Perform data processing on different types of first data to obtain standard data;

In this embodiment, due to different data types, the data magnitude of the extracted initial data may be greatly different. For example, the drug fee is 500, and the total cost is 1,000,000. There is a huge difference in the data magnitude between the two.

The server can perform data processing on initial data of different data levels through data processing methods of the same data level to obtain standard data of the same data level. For example, following the previous example, the drug fee and total cost are processed with the same data magnitude, and the drug fee and total cost with data magnitude between 0 and 100 are obtained, that is, the standard drug fee obtained is 0.05, and the standard total cost is is 100.

Specifically, data processing methods of the same data level can be selected based on different data types or different data levels. For example, methods such as square root, square, cube, exponential, logarithm, etc. can be selected, and this application does not limit this. .

508. Perform dimensionality reduction processing on the standard data to obtain target data with preset dimensions;

In this embodiment, the preset dimensions may be dimensions preset by the user on the server through the terminal according to subsequent data processing requirements. The data volume of the target data in the preset dimensions may be smaller than the data volume of the standard data. Nonlinear dimensionality reduction processing methods can include but are not limited to IsometricFeatureMapping (Isomap), Locally Linear Embedding (LLE), Modified Locally Linear Embedding (MLLE), Hessian Eigenmapping, Spectrum Embedding (Spectral Embedding), Local Tangent SpaceAlignment (LTSA), Multi-dimensional Scaling (MDS), t-distributedStochastic Neighbor Embedding (t-SNE) wait.

In practical applications, linear dimensionality reduction processing methods can also be used, which can include but are not limited to Principal Component Analysis (PCA), kernel PCA, and incremental principal component analysis (Incremental). PCA) etc. Specifically, the server can use the clustering characteristics of the data in the multi-dimensional Riemannian space according to the above method to map the multi-dimensional standard data to a low dimension, for example, to 2 dimensions, to obtain the target data.

In the above data dimensionality reduction processing method, the target medical features are obtained through the historical analysis model, and then the initial data corresponding to the target medical features in the multidimensional data are extracted, and the standard data is obtained after processing the data of the same data magnitude, and the standard numbers are Nonlinear dimensionality reduction processing is used to obtain target data with preset dimensions. The generated target data is generated based on multi-dimensional data and is related to the multi-dimensional data, so that the characteristics of the multi-dimensional data can be maintained, and subsequent data processing and analysis can be performed through the target data.

509. Perform pooling processing on the target data to obtain the medical feedforward vector of the target treatment data;

In this embodiment, the pooling analysis is based on the pooling proposed by the convolutional neural network. In this embodiment, maximum pooling (maxpooling) and average pooling (avg pooling) medical feature vectors are used to process, and a combination of maximum pooling is used. (max pooling) and mean pooling (avg pooling) perform pooling operations in parallel dual pooling layers to retain deeper semantic information of medical feature vectors.

510. According to the medical feedforward vector, perform cluster analysis on the medical information collection based on the preset cosine similarity algorithm, and obtain the medical service package corresponding to the disease.

Steps 501-503 and 510 in this embodiment are similar to steps 101-103 and 105 in the first embodiment, and will not be described again here.

The method for generating a service package based on patient data in the embodiment of the present application is described above. The device for generating a service package based on patient data in the embodiment of the present application is described below. Please refer to Figure 6. In the embodiment of the present application, the method of generating a service package based on patient data is described. A first embodiment of a data service package generating device includes:

The collection module 601 is used to collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

De-identification module 602 is used to de-identify the treatment data to obtain target treatment data;

The fusion module 603 is used to extract multiple key events in the target treatment data and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type;

Pooling module 604 is used to input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector;

The clustering module 605 is configured to perform cluster analysis on the medical information set based on the medical feedforward vector and a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease.

Please refer to Figure 7 , which is a second embodiment of a service package generation device based on patient data in the embodiment of this application. The service package generation device based on patient data specifically includes:

In this embodiment, the device for generating service packages based on patient data further includes:

The setting module 606 is used to pre-set diagnosis and treatment templates for different disease types;

The acquisition module 607 is used to obtain the time information corresponding to the treatment data; according to the disease type, obtain the diagnosis and treatment template corresponding to the treatment data;

The verification module 608 is used to verify the treatment data according to the diagnosis and treatment template, and determine whether the time sequence of the treatment data is correct.

In this embodiment, the de-identification module 602 is specifically used to:

Construct a treatment data query database based on the original case data;

Add random noise to the sensitive attribute fields in the data table of the treatment data query database according to the differential privacy algorithm;

According to the random noise, perform de-identification processing on the sensitive attribute fields in the original case data to obtain an identifier field;

The identifier field is encrypted to obtain target treatment data.

In this embodiment, the fusion module 603 is specifically used to:

Get a predefined key event collection;

Perform screening processing on the target treatment data to filter invalid medical data in the target treatment data to obtain valid medical data;

The effective medical data is extracted according to the key event set to obtain multiple key events in the target treatment data.

In this embodiment, the fusion module 603 is also specifically used to:

Determine the attribute characteristics corresponding to the key event and the weight value of the attribute characteristics;

The score data of the key event is determined based on the attribute characteristics and the weight value, and a medical information set of diseases corresponding to the disease type is obtained based on the score data.

In this embodiment, the pooling module 604 includes:

The feature extraction unit 6041 is used to extract features from the medical feature vector to obtain the target medical features of the medical feature vector;

The dimensionality reduction unit 6042 is used to perform dimensionality reduction processing on the target medical features to obtain target data of preset dimensions;

The pooling unit 6043 is used to perform pooling processing on the target data to obtain the medical feedforward vector of the target treatment data.

In this embodiment, the dimensionality reduction unit 6042 is specifically used to:

Obtain the frequency of the target medical feature;

extracting first data corresponding to the target medical feature in the target treatment data;

Perform data processing on the first data of different types to obtain standard data;

Perform dimensionality reduction processing on the standard data to obtain target data with preset dimensions.

The above Figure 6 and Figure 7 describe in detail the service package generation device based on patient data in the embodiment of the present application from the perspective of modular functional entities. The following is a detailed description of the service package based on patient data in the embodiment of the present application from the perspective of hardware processing. Generate a detailed description of the device.

Figure 8 is a schematic structural diagram of a service package generation device based on patient data provided by an embodiment of the present application. The service package generation device 800 based on patient data may vary greatly due to different configurations or performance, and may include one or One or more central processing units (CPU) 810 (e.g., one or more processors) and memory 820, one or more storage media 830 (e.g., one or more mass storage devices) storing applications 833 or data 832 ). Among them, the memory 820 and the storage medium 830 may be short-term storage or persistent storage. The program stored in the storage medium 830 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the service package generation device 800 based on patient data. Furthermore, the processor 810 may be configured to communicate with the storage medium 830 and execute a series of instruction operations in the storage medium 830 on the patient data-based service package generation device 800 to implement the disease-based service package provided by the above method embodiments. Steps of the service package generation method for suffering from data.

The service package generation device 800 based on patient data may also include one or more power supplies 840, one or more wired or wireless network interfaces 850, one or more input and output interfaces 860, and/or, one or more operating systems. 831, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art can understand that the structure of the service package generation device based on patient data shown in Figure 8 does not constitute a limitation on the service package generation device based on patient data provided in this application, and may include more or more features than those shown in the figure. Fewer parts, or combinations of certain parts, or different parts arrangements.

This application also provides a computer-readable storage medium. The computer-readable storage medium can be a non-volatile computer-readable storage medium. The computer-readable storage medium can also be a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are run on the computer, the computer is caused to execute the steps of the above-mentioned service package generation method based on patient data.

Those skilled in the field can clearly understand that for the convenience and simplicity of description, the specific working processes of the above-described systems, devices and units can be referred to the corresponding processes in the foregoing method embodiments, and will not be described again here.

If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or contributes to the existing technology, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium , including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned storage media include: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk and other media that can store program code. .

As mentioned above, the above embodiments are only used to illustrate the technical solution of the present application, but not to limit it. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still make the foregoing technical solutions. The technical solutions described in each embodiment may be modified, or some of the technical features may be equivalently replaced; however, these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions in each embodiment of the present application.

Claims

A method of generating service packages based on patient data, wherein the method of generating service packages based on patient data includes:

Collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

De-identify the treatment data to obtain target treatment data;

Extract multiple key events in the target treatment data, and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type;

Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector;

According to the medical feedforward vector, cluster analysis is performed on the medical information set based on a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease.
The method for generating service packages based on patient data according to claim 1, wherein original case data of similar diseases are collected from the preset medical information platform, and disease types and treatments in the original case data are extracted. After the data, it also includes:

Preset diagnosis and treatment templates for different disease types;

Obtain time information corresponding to the treatment data;

According to the disease type, obtain a diagnosis and treatment template corresponding to the treatment data;

The treatment data is verified according to the diagnosis and treatment template to determine whether the time sequence of the treatment data is correct.
The method for generating service packages based on patient data according to claim 1, wherein de-identifying the treatment data to obtain target treatment data includes:

Construct a treatment data query database based on the original case data;

Add random noise to the sensitive attribute fields in the data table of the treatment data query database according to the differential privacy algorithm;

According to the random noise, perform de-identification processing on the sensitive attribute fields in the original case data to obtain an identifier field;

The identifier field is encrypted to obtain target treatment data.
The method for generating service packages based on patient data according to claim 1, wherein the extracting a plurality of key events in the target treatment data includes:

Get a predefined key event collection;

Perform screening processing on the target treatment data to filter invalid medical data in the target treatment data to obtain valid medical data;

The effective medical data is extracted according to the key event set to obtain multiple key events in the target treatment data.
The method for generating service packages based on patient data according to claim 1, wherein said fusion processing of said key events to obtain a medical information set of diseases corresponding to said disease types includes:

Determine the attribute characteristics corresponding to the key event and the weight value of the attribute characteristics;

The score data of the key event is determined based on the attribute characteristics and the weight value, and a medical information set of diseases corresponding to the disease type is obtained based on the score data.
The method for generating service packages based on patient data according to claim 1, wherein said performing pooling analysis on the medical feature vector to obtain a medical feedforward vector includes:

Perform feature extraction on the medical feature vector to obtain the target medical features of the medical feature vector;

Perform dimensionality reduction processing on the target medical features to obtain target data with preset dimensions;

Pooling is performed on the target data to obtain a medical feedforward vector of the target treatment data.
The method for generating service packages based on patient data according to claim 6, wherein said performing dimensionality reduction processing on the target medical features to obtain target data of preset dimensions includes:

Obtain the frequency of the target medical feature;

extracting first data corresponding to the target medical feature in the target treatment data;

Perform data processing on the first data of different types to obtain standard data;

Perform dimensionality reduction processing on the standard data to obtain target data with preset dimensions.
A device for generating a service package based on patient data, wherein the device for generating a service package based on patient data includes:

The collection module is used to collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

A de-identification module, used to de-identify the treatment data to obtain target treatment data;

A fusion module, used to extract multiple key events in the target treatment data and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type;

The pooling module is used to input the medical information set into the preset Bilstm model for vector calculation, obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain the medical feedforward vector;

A clustering module is configured to perform cluster analysis on the medical information set based on the medical feedforward vector and a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease.
A device for generating a service package based on patient data, wherein the device for generating a service package based on patient data includes: a memory and at least one processor, instructions are stored in the memory, and the memory and the at least one Processors are interconnected by wires;

The at least one processor calls the instructions in the memory, so that the patient data-based service package generation device performs the steps of the patient data-based service package generation method as follows:

Collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

De-identify the treatment data to obtain target treatment data;

Extract multiple key events in the target treatment data, and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type;

Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector;

According to the medical feedforward vector, cluster analysis is performed on the medical information set based on a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease.
The device for generating service packages based on patient data according to claim 9, wherein the program for generating service packages based on patient data is executed by the processor to implement the collection of similar diseases from the preset medical information platform. After the steps of extracting the original case data and extracting the disease type and treatment data in the original case data, the following steps are also performed:

Preset diagnosis and treatment templates for different disease types;

Obtain time information corresponding to the treatment data;

According to the disease type, obtain a diagnosis and treatment template corresponding to the treatment data;

The treatment data is verified according to the diagnosis and treatment template to determine whether the time sequence of the treatment data is correct.
The device for generating a service package based on patient data according to claim 9, wherein the program for generating a service package based on patient data is executed by the processor to implement the de-identification processing of the treatment data, When obtaining target treatment data, the following steps are also performed:

Construct a treatment data query database based on the original case data;

Add random noise to the sensitive attribute fields in the data table of the treatment data query database according to the differential privacy algorithm;

According to the random noise, perform de-identification processing on the sensitive attribute fields in the original case data to obtain an identifier field;

The identifier field is encrypted to obtain target treatment data.
The device for generating a service package based on patient data according to claim 9, wherein the program for generating a service package based on patient data is executed by the processor to implement the extraction of a plurality of key points in the target treatment data. event steps, the following steps are also performed:

Get a predefined key event collection;

Perform screening processing on the target treatment data to filter invalid medical data in the target treatment data to obtain valid medical data;

The effective medical data is extracted according to the key event set to obtain multiple key events in the target treatment data.
The device for generating a service package based on patient data according to claim 9, wherein the program for generating a service package based on patient data is executed by the processor to implement the fusion processing of the key events to obtain When collecting the medical information of the disease corresponding to the disease type, the following steps are also performed:

Determine the attribute characteristics corresponding to the key event and the weight value of the attribute characteristics;

The score data of the key event is determined based on the attribute characteristics and the weight value, and a medical information set of diseases corresponding to the disease type is obtained based on the score data.
The device for generating a service package based on patient data according to claim 9, wherein the program for generating a service package based on patient data is executed by the processor to implement the pooling analysis of the medical feature vector, When obtaining the steps of medical feedforward vector, the following steps are also performed:

Perform feature extraction on the medical feature vector to obtain the target medical features of the medical feature vector;

Perform dimensionality reduction processing on the target medical features to obtain target data with preset dimensions;

Pooling is performed on the target data to obtain a medical feedforward vector of the target treatment data.
A computer-readable storage medium, a computer program stored on the computer-readable storage medium, wherein when the computer program is executed by a processor, the steps of the method for generating a service package based on patient data are implemented as follows:

Collect original case data of similar diseases from the preset medical information platform, and extract the disease type and treatment data in the original case data;

De-identify the treatment data to obtain target treatment data;

Extract multiple key events in the target treatment data, and perform fusion processing on the key events to obtain a medical information set of diseases corresponding to the disease type;

Input the medical information set into the preset Bilstm model for vector calculation to obtain the medical feature vector of the disease, and perform pooling analysis on the medical feature vector to obtain a medical feedforward vector;

According to the medical feedforward vector, cluster analysis is performed on the medical information set based on a preset cosine similarity algorithm to obtain a medical service package corresponding to the disease.
The computer-readable storage medium according to claim 15, wherein the computer program is executed by the processor to collect original case data of similar diseases from the preset medical information platform, and extract the original case data from the original case data. After the steps for disease type and treatment data, the following steps are also performed:

Preset diagnosis and treatment templates for different disease types;

Obtain time information corresponding to the treatment data;

According to the disease type, obtain a diagnosis and treatment template corresponding to the treatment data;

The treatment data is verified according to the diagnosis and treatment template to determine whether the time sequence of the treatment data is correct.
The computer-readable storage medium according to claim 15, wherein when the computer program is executed by the processor to de-identify the treatment data to obtain the target treatment data, the following steps are also performed:

Construct a treatment data query database based on the original case data;

Add random noise to the sensitive attribute fields in the data table of the treatment data query database according to the differential privacy algorithm;

According to the random noise, perform de-identification processing on the sensitive attribute fields in the original case data to obtain an identifier field;

The identifier field is encrypted to obtain target treatment data.
The computer-readable storage medium according to claim 15, wherein when the computer program is executed by the processor to extract a plurality of key events in the target treatment data, the following steps are also executed:

Get a predefined set of key events;

Perform screening processing on the target treatment data to filter invalid medical data in the target treatment data to obtain valid medical data;

The effective medical data is extracted according to the key event set to obtain multiple key events in the target treatment data.
The computer-readable storage medium according to claim 15, wherein when the computer program is executed by the processor, the step of fusing the key events to obtain a medical information set of diseases corresponding to the disease types is obtained. , also perform the following steps:

Determine the attribute characteristics corresponding to the key event and the weight value of the attribute characteristics;

The score data of the key event is determined based on the attribute characteristics and the weight value, and a medical information set of diseases corresponding to the disease type is obtained based on the score data.
The computer-readable storage medium according to claim 15, wherein when the computer program is executed by the processor to perform pooling analysis on the medical feature vector to obtain a medical feedforward vector, the following steps are also performed:

Perform feature extraction on the medical feature vector to obtain the target medical features of the medical feature vector;

Perform dimensionality reduction processing on the target medical features to obtain target data with preset dimensions;

Pooling is performed on the target data to obtain a medical feedforward vector of the target treatment data.