A medical data processing system and processing method

Abstract

This application relates to the field of data processing technology, and in particular to a medical data processing system and method for improving the accuracy of inference paths in accordance with medical logic. The main solution is as follows: Symptom data of the target patient is collected through a user client, and multimodal medical data of the target patient is collected through a medical data acquisition terminal; the symptom data and multimodal medical data of the target patient are input into an inference path generation model to obtain a set of inference paths, where each inference path consists of symptom nodes, disease nodes, and gene nodes; the path similarity probability corresponding to each inference path in the inference path set is determined based on a medical knowledge graph, which stores feature data corresponding to multiple medical paths; and an inference path is selected from the inference path set as the inference path for the target patient based on the path similarity probability and the inference path probability.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a medical data processing system and processing method. Background Technology

[0002] In current medical information systems, clinical data for various diseases is characterized by its massive volume, diverse sources, heterogeneity, and multimodality. How to efficiently manage and utilize this data is a critical issue that urgently needs to be addressed. Therefore, building an innovative and efficient intelligent disease pathway processing system is not only essential for improving the level of disease diagnosis and treatment, but also for significantly enhancing the quality of related clinical research. Summary of the Invention

[0003] In view of this, this application provides a medical data processing system and method to improve the accuracy of reasoning paths conforming to medical logic.

[0004] In a first aspect, embodiments of this application provide a medical data processing system, the system comprising: a medical data acquisition terminal, a user client, and a medical data processing server connected to the medical data acquisition terminal and the user client, the medical data processing server being used to perform the following steps:

[0005] The user client collects symptom data of the target patient, and the medical data collection terminal collects multimodal medical data of the target patient; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data.

[0006] The symptom data and multimodal medical data of the target patient are input into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0007] The path similarity probability corresponding to each reasoning path in the inference path set is determined based on the medical knowledge graph, and the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0008] Based on the path similarity probability and the inference path probability, an inference path is selected from the set of inference paths as the inference path for the target patient.

[0009] Secondly, embodiments of this application also provide a method for processing medical data, the method comprising:

[0010] The system collects symptom data of the target patient through a user client and multimodal medical data of the target patient through a medical data collection terminal; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data.

[0011] The symptom data and multimodal medical data of the target patient are input into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0012] The path similarity probability corresponding to each reasoning path in the inference path set is determined based on the medical knowledge graph, and the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0013] Based on the path similarity probability and the inference path probability, an inference path is selected from the set of inference paths as the inference path for the target patient.

[0014] Thirdly, embodiments of this application also provide an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device is running, the processor and the memory communicate via the bus, and the machine-readable instructions are executed by the processor to perform the steps of the medical data processing method of the second aspect.

[0015] Fourthly, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the medical data processing method described in the second aspect above.

[0016] This application provides a medical data processing system and method. First, it collects symptom data of a target patient through a user client and multimodal medical data of the target patient through a medical data acquisition terminal. This multimodal medical data includes at least gene testing data, medication data, and medical imaging data. Then, it inputs the target patient's symptom data and multimodal medical data into an inference path generation model to obtain an inference path set. This inference path set includes multiple inference paths and their corresponding probabilities. Each inference path is composed of symptom nodes, disease nodes, and gene nodes. Next, it determines the path similarity probability for each inference path in the inference path set based on a medical knowledge graph, which stores feature data corresponding to multiple medical paths. Finally, it selects one inference path from the inference path set as the target patient's inference path based on the path similarity probability and the inference path probability. Thus, this application achieves the determination of inference paths based on multimodal medical data. Since these inference paths include information such as symptoms, diseases, and genes, the resulting inference paths are more consistent with medical logic, thereby improving the effectiveness and accuracy of medical data processing.

[0017] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating a medical data processing method provided in an embodiment of this application is shown;

[0020] Figure 2 This paper shows a structural block diagram of a medical data processing server provided in an embodiment of this application;

[0021] Figure 3 A schematic diagram of a computer device provided in an embodiment of this application is shown. Detailed Implementation

[0022] The terms "first," "second," and "third," etc., used in this application specification, claims, and the aforementioned drawings are used to distinguish different objects, not to limit a specific order.

[0023] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0024] In the description of this application, unless otherwise stated, " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B can mean A or B. "And / or" in this application is merely a description of the relationship between the related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Furthermore, in the description of this application, unless otherwise stated, "multiple" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0025] In the embodiments of this application, at least one can also be described as one or more, and multiple can be two, three, four or more, and this application does not impose any restrictions.

[0026] This application provides a medical data processing system, which includes: a medical data acquisition terminal, a user client, and a medical data processing server connected to the medical data acquisition terminal and the user client. Figure 1 As shown, the medical data processing server in this embodiment is used to perform the following steps:

[0027] S10. Collect symptom data of the target patient through the user client and collect multimodal medical data of the target patient through the medical data collection terminal.

[0028] The user client is the client used by the patient, which can be a smartphone, computer, or other smart terminal. Through the user client, the patient can upload their symptoms. In this embodiment, the multimodal medical data includes at least gene testing data, medication data, and medical imaging data. Diseases are usually manifested through a series of symptoms; a disease may cause one or more symptoms, such as influenza, which may cause fever, cough, sore throat, etc. Symptom data can come from electronic health records or patient self-reports and is usually preprocessed into numerical vectors. Medication data refers to the medications recently taken by the target patient. Medical imaging data can specifically include X-ray images, magnetic resonance imaging, ultrasound imaging, etc., but this embodiment does not specifically limit these.

[0029] S20. Input the symptom data and multimodal medical data of the target patient into the inference path generation model to obtain the inference path set.

[0030] In this embodiment, the inference path generation model is a pre-trained neural network model. This model can predict a set of inference paths, which includes multiple inference paths and their corresponding probabilities. Each inference path is composed of symptom nodes, disease nodes, and gene nodes. The connecting edges between nodes in the inference path represent the relationships between them. The relationship between disease and symptom can be a manifestation relationship, a causal relationship, a diagnostic relationship, etc. Specifically, disease → symptom indicates that a disease causes a specific symptom (causal relationship), and symptom → disease indicates that the symptom points to a possible disease (diagnostic relationship). The relationship between disease and gene can be a genetic association, a pathogenic relationship, etc. Specifically, gene → disease indicates that a gene mutation directly causes a disease (monogenous disease), and disease → gene indicates that the disease involves multiple genes (polygenic disease). Disease indicates that a gene increases susceptibility to the disease. The relationship between symptoms and genes can be indirect, phenotypic, etc. Specifically, gene → symptom means that the gene causes symptoms by influencing physiological processes, and symptom → gene means that a specific symptom pattern indicates an abnormality in a specific gene.

[0031] For example, one reasoning path obtained from the reasoning path generation model is: fever + joint pain + rash (symptom node) → systemic lupus erythematosus (disease node) → HLA-DR gene (gene node).

[0032] In one optional embodiment provided in this application, the step of inputting the symptom data and multimodal medical data of the target patient into the inference path generation model to obtain an inference path set includes:

[0033] S201. Input the symptom data and multimodal medical data of the target patient into the inference path generation model, and obtain the symptom data vector and multimodal medical data vector of the target patient through the feature extraction layer in the inference path generation model.

[0034] In this embodiment, the symptom data vector can be represented as: ,in Let n represent the features of the i-th symptom (such as presence, intensity, or probability), and n be the number of symptoms. A multimodal medical data vector includes data vectors corresponding to gene testing data, medication data, and medical imaging data, respectively. A multimodal medical data vector can be represented as:

[0035]

[0036] Where B is a multimodal medical data vector, The data vector corresponding to the i-th gene detection data, The data vector corresponding to the i-th medication data. The data vector corresponding to the i-th medical image data, where m, p, and q are the quantities of gene testing data, medication data, and medical image data, respectively.

[0037] S202, the first prediction module in the reasoning path generation model predicts the disease label set based on the target patient's symptom data vector and multimodal medical data vector.

[0038] The disease label set includes multiple disease labels and their corresponding disease probabilities. Specifically, the first prediction module in the inference path generation model predicts the disease label set based on the target patient's symptom data vector and multimodal medical data vector, including: the first prediction module performs vector fusion on the target patient's symptom data vector and multimodal medical data vector to obtain a medical fusion data vector; the first prediction module uses a fully connected layer to map the medical fusion data vector to the disease space, and calculates the disease probability corresponding to each disease label using a softmax function; and adds disease labels with disease probabilities greater than a first probability value to the disease label set.

[0039] In this embodiment, symptom data vector A and multimodal medical data vector B can be concatenated to obtain a medical fusion data vector S. Specifically, this can be achieved using the formula... Obtain medical fusion data vector, S is the hidden layer feature vector, and S is the medical fusion data vector. This is the weight matrix. Here, is the bias vector, and ReLU is the activation function.

[0040] Then, according to the softmax function Calculate the disease probability corresponding to each disease label. This is the weight matrix. This is the bias vector. Specifically, the disease probabilities corresponding to each disease label obtained through the softmax function can be expressed by the formula... express, This represents the disease probability of the nth disease label obtained through medical fusion data vector prediction. This represents the nth disease label. Finally, disease labels with a probability greater than a first probability value are added to the disease label set. This first probability value can be set according to actual needs, such as 70%, 80%, etc. This embodiment does not impose a specific limitation on this.

[0041] S203. The second prediction module in the reasoning path generation model predicts the gene label set based on the disease label set, the target patient's symptom data vector, and the multimodal medical data vector.

[0042] The gene tag set includes gene tags and gene probabilities related to each disease tag in the disease tag set. Specifically, the second prediction module in the inference path generation model predicts the gene tag set based on the disease tag set, the target patient's symptom data vector, and the multimodal medical data vector. This includes: the second prediction module in the inference path generation model determining a disease-gene relationship set based on the disease tag set and the gene detection data vector in the target patient's multimodal medical data vector; the second prediction module in the inference path generation model determining a symptom-gene relationship set based on the target patient's symptom data vector and the gene detection data vector in the target patient's multimodal medical data vector; and obtaining the gene tag set through the union of the disease-gene relationship set and the genes in the symptom-gene relationship set.

[0043] It should be noted that many diseases are related to mutations or abnormal expression of specific genes. This relationship is a disease-gene relationship, meaning that a disease may be associated with one or more genes. For example, cystic fibrosis (a disease) is related to mutations in the CFTR gene (a gene). Genes are usually not directly associated with symptoms, but rather through a disease as an intermediary. That is, gene mutations lead to disease, and the disease then manifests as symptoms. Therefore, the relationship between symptoms and genes is indirect, but sometimes they can be directly related, for example, certain genes may directly lead to a predisposition to certain symptoms (such as pain sensitivity genes). Based on the above relationships, this embodiment needs to use the second prediction module in the inference path generation model to determine the disease-gene relationship set based on the disease label set and the gene detection data vector in the target patient's multimodal medical data vector; use the second prediction module in the inference path generation model to determine the symptom-gene relationship set based on the target patient's symptom data vector and the gene detection data vector in the target patient's multimodal medical data vector; and finally, obtain the gene label set by the union of the disease-gene relationship set and the genes in the symptom-gene relationship set.

[0044] S204. Obtain the inference path set through the symptom data vector, disease label set, and gene label set of the target patient.

[0045] Specifically, obtaining the inference path set through the target patient's symptom data vector, the disease tag set, and the gene tag set includes: determining a path candidate set through the target patient's symptom data vector, the disease tag set, and the gene tag set; calculating the base probability of each inference path in the path candidate set; calculating the corrected probability of each inference path in the path candidate set based on the confidence level of medical paths in the medical knowledge graph, wherein the corrected probability is the product of the confidence level and the base probability; and selecting an inference path from the path candidate set to add to the inference path set based on the corrected probability.

[0046] In this embodiment, the specific process of determining the path candidate set using the target patient's symptom data vector, the disease tag set, and the gene tag set can be as follows: First, significant symptoms (e.g., symptoms with intensity exceeding a threshold) are extracted from the symptom data vector; then, high-probability diseases (e.g., diseases with probability exceeding a threshold) are extracted from the disease tag set; next, high-probability genes (e.g., genes with probability exceeding a threshold) are extracted from the gene tag set; finally, all possible symptom-disease-gene combinations are generated to obtain the path candidate set. Furthermore, this embodiment can also perform preliminary screening based on some prior knowledge (such as the association between symptoms and diseases, and the association between diseases and genes) to reduce the number of paths in the path candidate set. For example, only triplets with symptom-disease association strength greater than a certain threshold and disease-gene association strength greater than a certain threshold are retained.

[0047] The calculation of the base probability of each inference path in the path candidate set includes: calculating the base probability of each inference path based on the symptom intensity value corresponding to the symptom of each inference path in the path candidate set, and the disease probability and gene probability corresponding to the disease label and gene label, respectively. Specifically, base probability = P_base(symptom s, disease d, gene g) = .

[0048] S30. Determine the path similarity probability corresponding to each reasoning path in the set of reasoning paths based on the medical knowledge graph.

[0049] The medical knowledge graph stores feature data corresponding to multiple medical paths. In an optional embodiment provided in this application, determining the path similarity probability corresponding to each inference path in the inference path set based on the medical knowledge graph includes: searching for the medical path corresponding to each inference path in the inference path set from the medical knowledge graph; and calculating the path similarity probability of the corresponding inference path based on the strength of the medical association between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path in the found medical path. The medical path records the strength of the medical association between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path.

[0050] The strength of the medical association between a symptom and a disease refers to a comprehensive measure of the frequency and specificity of a particular symptom in a given disease. It reflects the degree to which the symptom supports the diagnosis of the disease. The strength of the association can usually be converted into a value between 0 and 1. For example: 1.0 indicates a strong association (the symptom is highly specific and sensitive to the disease, such as Koplik spots for measles); 0.8 indicates a moderate association (the symptom is common in this disease but also seen in other diseases, such as chest pain for myocardial infarction); 0.5 indicates a weak association (the symptom may occur in this disease but is not a primary symptom, and is also common in other diseases, such as headache for hypertension); 0.2 indicates a very weak association (the relationship between the symptom and the disease is unclear, and it may be an occasional or non-specific manifestation); 0.0 indicates no association (the symptom is unrelated to the disease).

[0051] The pathogenicity association level between a disease and a gene refers to the strength and level of evidence of the causal relationship between a gene variant and the occurrence of a disease. It reflects the degree to which the gene variant contributes to the occurrence of the disease. Similarly, the pathogenicity association level can be converted into a number between 0 and 1. For example, 1.0 represents a known pathogenic gene (with clear mechanistic studies and repeating genetic evidence, such as the CFTR gene and cystic fibrosis); 0.8 represents a strong candidate gene (supported by multiple genetic studies, but the mechanism is not fully elucidated, or the evidence is slightly weaker), etc. The clinical support level of a complete pathway refers to the degree to which the entire chain of reasoning from symptoms to disease to gene is accepted and supported in clinical practice and medical research. It reflects the overall credibility and clinical applicability of the pathway. Similarly, clinical support can also be converted into a value between 0 and 1, for example: 1.0 represents high support (the pathway is widely accepted and included in clinical guidelines, with clear therapeutic significance, such as the HER2 gene and targeted therapy for breast cancer), etc.

[0052] Specifically, the step of calculating the path similarity probability of the corresponding inference path based on the strength of the medical association between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path in the found medical paths includes: weighting the calculation of the path similarity probability of the corresponding inference path based on the strength of the medical association between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path in the found medical paths.

[0053] S40. Select a reasoning path from the set of reasoning paths as the reasoning path for the target patient based on the path similarity probability and the reasoning path probability.

[0054] In this embodiment, selecting a reasoning path from the set of reasoning paths as the reasoning path for the target patient based on the path similarity probability and the reasoning path probability includes: performing a weighted calculation on the path similarity probability and the reasoning path probability to obtain the path probability of each reasoning path in the set of reasoning paths; and selecting the reasoning path with the highest path probability in the set of reasoning paths as the reasoning path for the target patient.

[0055] This application provides a medical data processing system. First, it collects symptom data of a target patient through a user client and multimodal medical data of the target patient through a medical data acquisition terminal. This multimodal medical data includes at least gene testing data, medication data, and medical imaging data. Then, it inputs the target patient's symptom data and multimodal medical data into an inference path generation model to obtain an inference path set. This inference path set includes multiple inference paths and their corresponding probabilities. Each inference path is composed of symptom nodes, disease nodes, and gene nodes. Next, it determines the path similarity probability for each inference path in the inference path set based on a medical knowledge graph that stores feature data corresponding to multiple medical paths. Finally, it selects one inference path from the inference path set as the target patient's inference path based on the path similarity probability and the inference path probability. Thus, this application achieves the determination of inference paths based on multimodal medical data. Since these inference paths include information such as symptoms, diseases, and genes, the resulting inference paths are more consistent with medical logic, thereby improving the effectiveness and accuracy of medical data processing.

[0056] When dividing each function into modules according to its corresponding function. Figure 2 This diagram illustrates a possible configuration of the medical data processing server described above and in the embodiments, such as... Figure 2 As shown, the medical data processing server may include:

[0057] The acquisition module 21 is used to acquire symptom data of the target patient through the user client and to acquire multimodal medical data of the target patient through the medical data acquisition terminal; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data;

[0058] Prediction module 22 is used to input the symptom data and multimodal medical data of the target patient into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0059] The determination module 23 is used to determine the path similarity probability corresponding to each reasoning path in the inference path set based on the medical knowledge graph, wherein the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0060] Selection module 24 is used to select one inference path from the set of inference paths as the inference path for the target patient based on the path similarity probability and the inference path probability.

[0061] In an optional embodiment provided by the present invention, the prediction module 22 is specifically used for:

[0062] The symptom data and multimodal medical data of the target patient are input into the inference path generation model, and the symptom data vector and multimodal medical data vector of the target patient are obtained through the feature extraction layer in the inference path generation model.

[0063] The first prediction module in the inference path generation model predicts a disease label set based on the symptom data vector and multimodal medical data vector of the target patient. The disease label set includes multiple disease labels and their corresponding disease probabilities.

[0064] The second prediction module in the inference path generation model predicts a set of gene tags based on the disease tag set, the symptom data vector of the target patient, and the multimodal medical data vector. The set of gene tags includes gene tags and gene probabilities related to each disease tag in the disease tag set.

[0065] The inference path set is obtained by using the symptom data vector of the target patient, the disease tag set, and the gene tag set.

[0066] In an optional embodiment provided by the present invention, the prediction module 22 is specifically used for:

[0067] The first prediction module performs vector fusion on the symptom data vector and multimodal medical data vector of the target patient to obtain a medical fusion data vector;

[0068] The first prediction module uses a fully connected layer to map the medical fusion data vector to the disease space, and calculates the disease probability corresponding to each disease label through the softmax function;

[0069] Add disease labels whose disease probability is greater than the first probability value to the disease label set.

[0070] In an optional embodiment provided by the present invention, the prediction module 22 is specifically used for:

[0071] The second prediction module in the inference path generation model determines the disease-gene relationship set based on the disease label set and the gene detection data vector in the multimodal medical data vector of the target patient.

[0072] The second prediction module in the inference path generation model determines the set of symptom-gene relationships based on the symptom data vector of the target patient and the gene detection data vector in the multimodal medical data vector of the target patient.

[0073] A gene tag set is obtained by the union of genes in the disease-gene relationship set and the symptom-gene relationship set.

[0074] In an optional embodiment provided by the present invention, the prediction module 22 is specifically used for:

[0075] A path candidate set is determined using the symptom data vector of the target patient, the disease tag set, and the gene tag set.

[0076] Calculate the base probability of each inference path in the path candidate set;

[0077] The correction probability of each inference path in the candidate path set is calculated based on the confidence level of the medical path in the medical knowledge graph. The correction probability is the product of the confidence level and the base probability.

[0078] Based on the correction probability, a reasoning path is selected from the path candidate set and added to the reasoning path set.

[0079] In an optional embodiment provided by the present invention, the prediction module 22 is specifically used for:

[0080] The base probability of each inference path is calculated based on the symptom intensity value corresponding to the symptom of each inference path in the candidate path set, as well as the disease probability and gene probability corresponding to the disease label and gene label, respectively.

[0081] In an optional embodiment provided by the present invention, the determining module 23 is specifically used for:

[0082] Find the medical path corresponding to each reasoning path in the set of reasoning paths from the medical knowledge graph;

[0083] The path similarity probability of the corresponding inference path is calculated based on the strength of the medical association between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path in the found medical pathways.

[0084] In an optional embodiment provided by the present invention, the determining module 23 is specifically used for:

[0085] The path similarity probability of the corresponding inference path is obtained by weighting the medical association strength between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support of the complete path in the found medical paths.

[0086] In an optional embodiment provided by the present invention, the selection module 24 is specifically used for:

[0087] The path probability of each inference path in the inference path set is obtained by weighting the path similarity probability and the inference path probability.

[0088] The inference path with the highest probability in the inference path set is taken as the inference path for the target patient.

[0089] For specific limitations regarding the system, please refer to the limitations on medical data processing methods described above, which will not be repeated here. Each module in the aforementioned device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of the computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to each module.

[0090] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 3 As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The network interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for processing medical data.

[0091] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps:

[0092] The system collects symptom data of the target patient through a user client and multimodal medical data of the target patient through a medical data collection terminal; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data.

[0093] The symptom data and multimodal medical data of the target patient are input into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0094] The path similarity probability corresponding to each reasoning path in the inference path set is determined based on the medical knowledge graph, and the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0095] Based on the path similarity probability and the inference path probability, an inference path is selected from the set of inference paths as the inference path for the target patient.

[0096] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor:

[0097] The system collects symptom data of the target patient through a user client and multimodal medical data of the target patient through a medical data collection terminal; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data.

[0098] The symptom data and multimodal medical data of the target patient are input into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0099] The path similarity probability corresponding to each reasoning path in the inference path set is determined based on the medical knowledge graph, and the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0100] Based on the path similarity probability and the inference path probability, an inference path is selected from the set of inference paths as the inference path for the target patient.

[0101] In one embodiment, a computer program product is provided, the computer program product comprising a computer program that is executed by a processor to perform the following steps:

[0102] The system collects symptom data of the target patient through a user client and multimodal medical data of the target patient through a medical data collection terminal; the multimodal medical data includes at least gene testing data, medication data, and medical imaging data.

[0103] The symptom data and multimodal medical data of the target patient are input into the inference path generation model to obtain an inference path set. The inference path set includes multiple inference paths and corresponding inference path probabilities. The inference path is a path composed of symptom nodes, disease nodes, and gene nodes.

[0104] The path similarity probability corresponding to each reasoning path in the inference path set is determined based on the medical knowledge graph, and the medical knowledge graph stores feature data corresponding to multiple medical paths respectively;

[0105] Based on the path similarity probability and the inference path probability, an inference path is selected from the set of inference paths as the inference path for the target patient.

[0106] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0107] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0108] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A medical data processing system, characterized by, The system comprises a medical data acquisition terminal, a user client, and a medical data processing server connected with the medical data acquisition terminal and the user client, and the medical data processing server is configured to perform the following steps: Collecting symptom data of a target patient through the user client and collecting multi-modal medical data of the target patient through the medical data acquisition terminal; the multi-modal medical data at least includes genetic testing data, medication data, and medical image data; Inputting the symptom data and the multi-modal medical data of the target patient into an inference path generation model to obtain an inference path set, the inference path set including a plurality of inference paths and corresponding inference path probabilities, the inference path being a path composed of a symptom node, a disease node, and a gene node, and the connection edges between the nodes in the inference path being used to represent the relationship between the nodes; the inference path generation model being a pre-trained neural network model; Determining a path similarity probability corresponding to each inference path in the inference path set according to a medical knowledge graph, the medical knowledge graph storing feature data corresponding to a plurality of medical paths; Selecting an inference path from the inference path set as the inference path of the target patient according to the path similarity probability and the inference path probability; The method of determining a path similarity probability corresponding to each inference path in the inference path set according to a medical knowledge graph comprises: Finding a medical path corresponding to each inference path in the inference path set from the medical knowledge graph; Performing weighted calculation on the medical association strength between symptoms and diseases in the found medical path, the pathogenic association level between diseases and genes, and the clinical support degree of the complete path to obtain the path similarity probability of the corresponding inference path.

2. The system of claim 1, wherein, The method of inputting the symptom data and the multi-modal medical data of the target patient into an inference path generation model to obtain an inference path set comprises: Inputting the symptom data and the multi-modal medical data of the target patient into the inference path generation model to obtain a symptom data vector and a multi-modal medical data vector through a feature extraction layer in the inference path generation model; A first prediction module in the inference path generation model predicts a disease label set according to the symptom data vector and the multi-modal medical data vector of the target patient, the disease label set including a plurality of disease labels and corresponding disease probabilities; A second prediction module in the inference path generation model predicts a gene label set according to the disease label set, the symptom data vector, and the multi-modal medical data vector of the target patient, the gene label set including gene labels and gene probabilities related to each disease label in the disease label set; An inference path set is obtained through the symptom data vector, the disease label set, and the gene label set of the target patient.

3. The system of claim 2, wherein, The method of predicting a disease label set by a first prediction module in the inference path generation model according to the symptom data vector and the multi-modal medical data vector of the target patient comprises: The first prediction module performs vector fusion on the symptom data vector and the multi-modal medical data vector of the target patient to obtain a medical fusion data vector; The first prediction module uses a full connection layer to map the medical fusion data vector to a disease space, and calculates a disease probability corresponding to each disease label by using a softmax function; A disease label with a disease probability greater than a first probability value is added to the disease label set.

4. The system of claim 3, wherein, The second prediction module in the reasoning path generation model predicts a gene label set according to the disease label set, the symptom data vector and the multi-modal medical data vector of the target patient, including: The second prediction module in the reasoning path generation model determines a disease-gene relationship set according to the disease label set and a gene detection data vector in the multi-modal medical data vector of the target patient; The second prediction module in the reasoning path generation model determines a symptom-gene relationship set according to the symptom data vector of the target patient and the gene detection data vector in the multi-modal medical data vector of the target patient; The gene label set is obtained by taking the union of genes in the disease-gene relationship set and the symptom-gene relationship set.

5. The system of claim 2, wherein, The reasoning path set is obtained by the symptom data vector of the target patient, the disease label set and the gene label set, including: A path candidate set is determined by the symptom data vector of the target patient, the disease label set and the gene label set; A base probability of each reasoning path in the path candidate set is calculated; A correction probability of each reasoning path in the path candidate set is calculated according to a confidence of a medical path in a medical knowledge graph, and the correction probability is a product of the confidence and the base probability; A reasoning path is selected from the path candidate set according to the correction probability and added to the reasoning path set.

6. The system of claim 5, wherein, The base probability of each reasoning path in the path candidate set is calculated, including: The base probability of each reasoning path is calculated according to a symptom intensity value corresponding to a symptom of each reasoning path in the path candidate set, and a disease probability and a gene probability corresponding to a disease label and a gene label, respectively.

7. The system of claim 1, wherein, The reasoning path is selected from the reasoning path set according to the path similarity probability and the reasoning path probability, including: A path probability of each reasoning path in the reasoning path set is calculated by weighting the path similarity probability and the reasoning path probability; The reasoning path with the highest path probability in the reasoning path set is selected as the reasoning path of the target patient.

8. A method of processing medical data, characterized by, The method is applied to a medical data processing server in a medical data processing system according to any one of claims 1-7, and the method includes: Symptom data of a target patient is collected by a user client, and multi-modal medical data of the target patient is collected by a medical data collection terminal; the multi-modal medical data at least includes gene detection data, medication data and medical image data; inputting symptom data and multi-modal medical data of the target patient into an inference path generation model to obtain an inference path set, the inference path set including a plurality of inference paths and corresponding inference path probabilities, the inference path being a path composed of symptom nodes, disease nodes and gene nodes, and the connection edges between the nodes in the inference path being used to represent the relationships between the nodes; the inference path generation model being a pre-trained neural network model; finding a medical path corresponding to each inference path in the inference path set from a medical knowledge graph; and performing weighted calculation on the medical association strength between symptoms and diseases, the pathogenic association level between diseases and genes, and the clinical support degree of the complete path in the found medical path to obtain a path similarity probability of the corresponding inference path, the medical knowledge graph storing feature data corresponding to a plurality of medical paths; selecting an inference path from the inference path set as the inference path of the target patient according to the path similarity probability and the inference path probability.

Images