Training method and system of disease course record generation model

Abstract

The invention provides a training method and system for a disease course record generation model, and the method comprises the steps: obtaining an initial patient question, obtaining an initial disease course record corresponding to the initial patient question and determined by simulating the diagnosis thought of a doctor, and carrying out the training of the initial disease course record from a pre-constructed standardized disease course record database, a disease course record template corresponding to the initial disease course record is obtained through retrieval, the standardized disease course record database comprises templates corresponding to the disease course records and determined on the basis of diagnosis ideas of doctors, the initial disease course record is corrected according to the disease course record template, and a target disease course record is obtained. And performing fine adjustment on the basic network model according to the target disease course record to obtain a disease course record generation model. Therefore, the obtained target disease course record better conforms to the diagnosis thought of a doctor. Therefore, the disease course record generation model obtained based on training of the target disease course record can accurately and reliably simulate the diagnosis thought of a doctor, and the corresponding disease course record is generated and output.

Description

Technical Field

[0001] This specification relates to the field of artificial intelligence technology, and in particular to a training method and system for a medical record generation model. Background Art

[0002] The medical record can be understood as a continuous record of the patient's condition and treatment process following the hospitalization record.

[0003] With the development of artificial intelligence technology, it has been widely used in various fields, such as the medical field. In today's medical field, the huge and complex patient diagnosis and treatment data are facing major management and application challenges.

[0004] Existing large medical models can often only directly output conclusions about patient problems, and it is difficult to conduct in-depth reasoning through diagnostic logic like a doctor.

[0005] It is worth noting that the content of the above-mentioned related technologies is only information known to the inventor personally, and does not mean that the above-mentioned information has entered the public domain before the filing date of this specification, nor does it mean that it can become the prior art of this specification. Summary of the Invention

[0006] This specification provides a training method and system for a medical record generation model to avoid at least one of the above-mentioned technical problems.

[0007] In a first aspect, this specification provides a method for training a process record generation model, comprising:

[0008] Obtaining an initial patient problem and obtaining an initial medical record corresponding to the initial patient problem determined by simulating a doctor's diagnostic thinking, wherein the initial medical record is used to represent a diagnostic evidence chain corresponding to the diagnostic thinking;

[0009] Retrieving a medical record template corresponding to the initial medical record from a pre-constructed standardized medical record database, wherein the standardized medical record database includes a template corresponding to each medical record determined based on the doctor's diagnostic thinking;

[0010] Correcting the initial medical record according to the medical record template to obtain a target medical record; and

[0011] The basic network model is fine-tuned according to the target medical record to obtain a medical record generation model, wherein the medical record generation model is used to simulate the doctor's diagnostic thinking and generate and output the medical record corresponding to the target patient's problem.

[0012] In a second aspect, this specification provides a method for generating a medical record, comprising:

[0013] Obtain targeted patient questions;

[0014] Inputting the target patient's problem into a medical record generation model, simulating the doctor's diagnostic thinking based on the medical record generation model, and generating and outputting a medical record corresponding to the target patient's problem;

[0015] The medical record generation model is obtained by training based on the training method described in the first aspect.

[0016] In a third aspect, this specification provides a training system for a medical record generation model, comprising:

[0017] At least one storage medium storing at least one instruction set for training a medical record generation model;

[0018] At least one processor is communicatively connected to the at least one storage medium, wherein when the at least one processor is running, it reads the at least one instruction set and executes the training method as described in the first aspect according to the instructions of the at least one instruction set.

[0019] In a fourth aspect, this specification provides a system for generating medical records, comprising:

[0020] At least one storage medium storing at least one instruction set for training a medical record generation model;

[0021] At least one processor is communicatively connected to the at least one storage medium, wherein when the at least one processor is running, it reads the at least one instruction set and executes the generation method as described in the second aspect according to the instructions of the at least one instruction set.

[0022] In a fifth aspect, this specification provides a computer-readable non-temporary storage medium, wherein the computer-readable non-temporary storage medium stores at least one instruction set, and the at least one instruction set is executed by at least one processor to implement the method described in the first aspect or the second aspect.

[0023] As can be seen from the above technical solutions, the training method and system for the medical record generation model provided in this specification corrects the initial medical record based on the medical record template in the standardized medical record database, so that the resulting target medical record is more consistent with the doctor's diagnostic thinking (i.e., diagnostic logic). As a result, the medical record generation model trained based on the target medical record can imitate the doctor's diagnostic thinking and generate and output the corresponding medical record.

[0024] Other functions of the training method and system for the medical record generation model provided in this specification will be partially listed in the following description. The creative aspects of the training method and system for the medical record generation model provided in this specification can be fully explained by practicing or using the methods, devices and combinations described in the following detailed examples. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of this specification, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 A schematic diagram of an application scenario of the training method for the medical record generation model provided in the embodiments of this specification;

[0027] Figure 2 A schematic diagram of the structure of a training system for a medical record generation model provided in an embodiment of this specification;

[0028] Figure 3 A flowchart of a method for training a medical record generation model provided in one embodiment of this specification;

[0029] Figure 4 A flowchart of a method for training a medical record generation model provided in another embodiment of this specification;

[0030] Figure 5 A schematic diagram illustrating the principle of a training method for a medical record generation model provided in an embodiment of this specification;

[0031] Figure 6 A schematic diagram of obtaining quality assessment information based on an interactive method according to an embodiment of this specification;

[0032] Figure 7 A flowchart of a method for generating a medical record according to an embodiment of this specification;

[0033] Figure 8 This is a schematic diagram of the principle of the method for generating medical records provided in the embodiments of this specification. DETAILED DESCRIPTION

[0034] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with this specification. Rather, they are merely examples of apparatus and methods consistent with certain aspects of this specification, as detailed in the appended claims.

[0035] It should be understood that the terms "including" and "having" and any variations thereof in the embodiments of this specification are intended to cover but not exclude inclusion. For example, a product or device that includes a series of components is not necessarily limited to those components explicitly listed, but may include other components that are not explicitly listed or are inherent to these products or devices.

[0036] In the embodiments of this specification, the term "and / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0037] The term "plurality" in the embodiments of this specification refers to two or more than two, and other quantifiers are similar to it.

[0038] The terms "first," "second," "third," "initial," "target," "sample," "template," etc., in this specification are used to distinguish similar or similar objects or entities and are not necessarily intended to limit a particular order or precedence, unless otherwise indicated. It should be understood that the terms used in this manner are interchangeable where appropriate, for example, enabling implementation in an order other than that shown or described in the embodiments of this specification.

[0039] The term "unit / module" as used in this specification refers to any known or later developed hardware, software, firmware, artificial intelligence, fuzzy logic or combination of hardware and / or software code that is capable of performing the functions associated with that element.

[0040] In order to avoid at least one of the technical problems mentioned in the above background technology, this specification proposes a technical conception obtained through creative labor: First, by simulating the doctor's diagnostic thinking, the initial medical record corresponding to the initial patient problem is determined. Then, a medical record template corresponding to the initial medical record is obtained from the templates corresponding to each medical record determined based on the doctor's diagnostic thinking in the real scene. Next, the initial medical record is corrected based on the medical record template to obtain a corrected target medical record, so that the target medical record is relatively more in line with the doctor's diagnostic thinking. Finally, the basic network model is fine-tuned based on the target medical record to obtain a medical record generation model for accurately simulating the doctor's diagnostic thinking, generating and outputting the medical record corresponding to the target patient problem.

[0041] The technical solution provided in this specification is based on the aforementioned technical concept. Based on the description of the aforementioned technical concept, it can be seen that in the technical solution provided in this specification, the initial medical record is corrected using a medical record template so that the resulting target medical record is more consistent with the doctor's diagnostic thinking. As a result, the medical record generation model trained based on the target medical record can accurately and reliably mimic the doctor's diagnostic thinking and generate and output the corresponding medical record.

[0042] Accordingly, when determining the corresponding medical records based on the medical record generation model, it is more consistent with the doctor's way of thinking, effectively assisting doctors in diagnosis. Furthermore, the reasoning process can be made transparent, increasing the trustworthiness of medical records, and it has a high degree of customization and strong optimizability.

[0043] To facilitate readers' understanding of this specification, the application scenarios of the training method for the medical record generation model provided in this specification (hereinafter referred to as the training method) are now introduced.

[0044] The training method provided in this specification is suitable for scenarios where a model for generating medical records needs to be trained. For example, the technical solution provided in this specification can be applied to scenarios such as automated medical record writing, standardized medical record templates, intelligent auxiliary diagnosis, integration of telemedicine and electronic health records (EHRs), personalized health management, scientific research, and education.

[0045] Take the above scenario of automated medical record writing as an example:

[0046] Using the training method provided in this manual, a medical record generation model can be trained to simulate a doctor's diagnostic thinking. Accordingly, the medical record generation model can receive questions input by patients via voice or text, and automatically generate and output corresponding medical records based on these questions. This not only saves doctors time in writing medical records, but also improves their accuracy and completeness.

[0047] Take the above scenario of intelligent assisted diagnosis as an example:

[0048] Using the training methods provided in this manual, a medical record generation model can be trained to simulate a doctor's diagnostic thinking. Accordingly, the medical record generation model can receive questions input by the patient via voice or text, and automatically generate and output corresponding medical records based on the questions. Based on these medical records, doctors can quickly review the patient's medical history, identify potential health issues, and make more accurate diagnoses.

[0049] For descriptions of how the training method provided in this manual may be applied to other scenarios, please refer to the above examples, which will not be listed here one by one.

[0050] Figure 1 The following is a schematic diagram of an application scenario of the training method of the embodiment of this specification, wherein the training method of this specification can be applied to the following Figure 1 The scene shown is 100. Figure 1 As shown, scenario 100 may include a target user 101 , a client 102 , a server 103 , and a network 104 .

[0051] The target user 101 may be a user who triggers the training of the medical record generation model. For example, the target user 101 may perform a target operation on the client 102 to trigger the training of the medical record generation model.

[0052] The client 102 may be an electronic device that provides interactive functions to the target user 101. For example, the client 102 may provide an interactive interface to the target user 101, and the target user 101 may perform interactive operations in the interactive page. In some embodiments, the client 102 executes the training method described in this specification in response to detecting the target operation triggered by the target user 101 to train the medical record generation model. At this time, the client 102 may store data or instructions for executing the training method described in this specification, and may execute or be used to execute the data or instructions. In some embodiments, the client 102 may include a hardware device with data information processing capabilities and the necessary programs required to drive the hardware device to work, so as to execute the training method described in this specification.

[0053] In some embodiments, client 102 may include a mobile device, a tablet computer, a laptop computer, a built-in device in a motor vehicle, or the like, or any combination thereof. In some embodiments, the mobile device may include a smart home device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the smart home device may include a smart television, a desktop computer, or the like, or any combination thereof. In some embodiments, the smart mobile device may include a smartphone, a personal digital assistant, a gaming device, a navigation device, or the like, or any combination thereof. In some embodiments, the built-in device in a motor vehicle may include an onboard computer, an onboard television, or the like.

[0054] In some embodiments, the client 102 may be installed with one or more applications (APPs). APPs can provide the target user 101 with the ability and interface to interact with the outside world via the network 104. APPs include, but are not limited to, web browser APPs, search APPs, chat APPs, shopping APPs, video APPs, financial management APPs, instant messaging tools, email clients, social networking platform software, and the like.

[0055] like Figure 1 As shown, the client 102 can be in communication connection with the server 103. The server 103 can be in communication connection with one client 102 or with multiple clients 102. In some embodiments, the client 102 can interact with the server 103 via the network 104 to receive or send messages, etc.

[0056] The server 103 may be a server that provides various services. For example, the server 103 may be a cloud server or a local server. The server 103 may be connected to a client 102 and receive data sent by the client 102, or may be connected to multiple clients 102 and receive data sent by each client 102.

[0057] In some embodiments, the training methods described herein can be executed on server 103. In this case, server 103 can store data or instructions for executing the training methods described herein and can execute or be used to execute the data or instructions. Server 103 can include hardware devices capable of data information processing and the necessary programs to operate the hardware devices.

[0058] The network 104 is a medium for providing a communication connection between the client 102 and the server 103. The network 104 can facilitate the exchange of information or data. Figure 1As shown, the client 102 and the server 103 can be connected to the network 104 respectively, and transmit information or data to each other through the network 104.

[0059] In some embodiments, the network 104 can be any type of wired or wireless network, or a combination thereof. For example, the network 104 can include a cable network, a wired network, a fiber optic network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a public switched telephone network (PSTN), a Bluetooth™ network, a ZigBee™ network, a near field communication (NFC) network, or the like.

[0060] In some embodiments, network 104 may include one or more network access points. For example, network 104 may include a wired or wireless network access point, such as a base station or an Internet exchange point, through which one or more components of client 102 and server 103 can connect to network 104 to exchange data or information.

[0061] It is worth mentioning that Figure 1 The number of clients 102, servers 103, and networks 104 in the examples is merely illustrative. Any number of clients 102, servers 103, and networks 104 may be used as needed. Furthermore, the training method provided herein may be executed entirely on the client 102, entirely on the server 103, or partially on the client 102 and partially on the server 103.

[0062] That is to say, Figure 1 and targeting Figure 1 The above description is only used to illustrate the application scenarios to which the training method of this specification may be applicable, and should not be understood as a limitation on the application scenarios.

[0063] Figure 2The hardware structure diagram of a training system 200 for a medical record generation model provided according to an embodiment of the present specification (hereinafter referred to as the training system) is shown. The training system 200 can execute the training method described in this specification. The training method is introduced in other parts of this specification. When the training method is executed on the client 102, the training system 200 can be the client 102. When the training method is executed on the server 103, the training system 200 can be the server 103. When the training method is partially executed on the client 102 and partially executed on the server 103, the training system 200 can be a system including the client 102 and the server 103.

[0064] like Figure 2 As shown, training system 200 may include at least one storage medium 203 and at least one processor 202. In some embodiments, training system 200 may further include a communication port 204 and an internal communication bus 201. Training system 200 may further include an I / O component 205.

[0065] The internal communication bus 201 can connect different system components. For example, the internal communication bus 201 can connect the storage medium 203, the processor 202, the communication port 204 and the I / O component 205.

[0066] I / O components 205 support input / output between training system 200 and other components.

[0067] Communication port 204 is used for data communication between training system 200 and the outside world. For example, communication port 204 can be used for data communication between training system 200 and network 104. Communication port 204 can be a wired communication port or a wireless communication port.

[0068] Storage medium 203 may include a data storage device. The data storage device may be a non-transitory storage medium or a temporary storage medium. For example, the data storage device may include one or more of a disk 2031, a read-only storage medium (ROM) 2032, or a random access storage medium (RAM) 2033. Storage medium 203 also includes at least one instruction set stored in the data storage device. The instruction set includes computer program code, which may include programs, routines, objects, components, data structures, processes, modules, etc. that execute the training methods provided in this specification.

[0069] At least one processor 202 can be communicatively connected to at least one storage medium 203. The at least one processor 202 is configured to execute the at least one instruction set described above. When the training system 200 is running, the at least one processor 202 reads the at least one instruction set and, according to the instructions of the at least one instruction set, executes the training method provided herein. The processor 202 can perform all steps included in the training method. The processor 202 can be in the form of one or more processors. In some embodiments, the processor 202 can include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a central processing unit (CPU), a graphics processing unit (GPU), a physical processing unit (PPU), a microcontroller unit, a digital signal processor (DSP), a field programmable gate array (FPGA), an advanced RISC machine (ARM), a programmable logic device (PLD), any circuit or processor capable of performing one or more functions, or any combination thereof.

[0070] For illustrative purposes only, only one processor 202 is shown in the training system 200 in the accompanying drawings. However, it should be noted that the training system 200 described herein may also include multiple processors. Therefore, the operations and / or method steps disclosed herein may be performed by a single processor or jointly by multiple processors. For example, if the training system 200 is described herein as processor 202 performing steps A and B, it should be understood that steps A and B may also be performed jointly or separately by two different processors 202 (e.g., a first processor performing step A and a second processor performing step B, or a first and a second processor performing steps A and B together).

[0071] See also Figure 3 , Figure 3 This is a flow chart of a method for training a medical record generation model according to one embodiment of this specification. Figure 3 The training method shown may be executed by a training system. For a description of the training system, please refer to the above examples and will not be repeated here.

[0072] like Figure 3 As shown, the method includes the following S301 to S304:

[0073] S301: Obtain an initial patient problem, and obtain an initial medical record corresponding to the initial patient problem determined by simulating a doctor's diagnostic thinking. The initial medical record is used to represent a diagnostic evidence chain corresponding to the diagnostic thinking.

[0074] Initial patient questions can be understood as questions initiated by patients when training the medical record generation model. For example, combined with the above description of the application scenario, the training system can obtain questions initiated by patients in voice or text format.

[0075] The initial medical record can be understood as a medical record that corresponds to the initial patient's problem and aligns with the physician's diagnostic thinking. The initial medical record can represent a chain of diagnostic evidence that aligns with the physician's diagnostic thinking. For example, the initial medical record simulates the physician's diagnostic thinking, analyzing symptoms, signs, and laboratory test information for the initial patient's problem to form a clear chain of diagnostic evidence.

[0076] For example, the contents of the initial medical record include the case characteristics corresponding to the initial patient problems, the proposed diagnosis discussion (diagnostic basis and differential diagnosis), the diagnosis and treatment plan, etc. Among them, the case characteristics can be understood as the case characteristics determined after a comprehensive analysis, summary and organization of the medical history, physical examination and auxiliary examinations. Such as positive findings and negative symptoms and signs with differential diagnostic significance. The proposed diagnosis discussion (diagnostic basis and differential diagnosis) can be understood as proposing a preliminary diagnosis and diagnostic basis based on the characteristics of the case. Write down the differential diagnosis and analyze it for unclear diagnoses, and analyze the next step of diagnosis and treatment measures. The diagnosis and treatment plan can be understood as the specific examination and treatment measures determined.

[0077] S302: Retrieve a medical record template corresponding to the initial medical record from a pre-built standardized medical record database, wherein the standardized medical record database includes a template corresponding to each medical record determined based on the doctor's diagnosis idea.

[0078] For example, the training system and other systems may pre-build a standardized medical record database to store templates corresponding to each medical record in the standardized medical record database, and each template is determined by the doctor's diagnostic thinking.

[0079] For example, a training system can construct a template for each disease's medical history based on a doctor's diagnostic thinking. After obtaining the initial medical history, the training system can search the standardized medical history data to obtain a template corresponding to the initial medical history.

[0080] S303: Correct the initial medical record according to the medical record template to obtain a target medical record.

[0081] In contrast, the medical record template is based on the doctor's diagnostic thinking, and the initial medical record is created by simulating the doctor's diagnostic thinking. The medical record template can be understood as the standard for medical records. Therefore, the medical record template more closely represents the doctor's diagnostic thinking.

[0082] Accordingly, the training system can modify the initial medical record of the simulated doctor's diagnostic thinking based on a medical record template that is more in line with the doctor's diagnostic thinking, so as to obtain a target medical record that is more in line with the doctor's diagnostic thinking compared to the initial medical record.

[0083] In some embodiments, each template in the standardized medical record database is generated by doctors by labeling the corresponding diseases based on diagnostic ideas.

[0084] For example, for a certain disease, a doctor can label the disease based on his or her diagnostic ideas, thereby obtaining a template corresponding to the disease.

[0085] Accordingly, the training system can create a standardized medical record database that includes templates corresponding to each disease.

[0086] In contrast, the training system generates a standardized medical record database by annotating the corresponding templates for each disease based on doctors' diagnostic thinking. This allows the standardized medical record database to accurately and reliably represent doctors' diagnostic thinking, ensuring that each template has detailed diagnostic evidence and thought processes. This improves the accuracy and practicality of the standardized medical record database. Furthermore, through doctor annotation, not only does the generation system's reasoning capabilities continue to improve, but it also enables iterative improvements through human-computer collaboration.

[0087] S304: Fine-tune the basic network model according to the target medical record to obtain a medical record generation model, wherein the medical record generation model is used to simulate the doctor's diagnostic thinking and generate and output the medical record corresponding to the target patient's problem.

[0088] For example, when a target medical record is obtained that is relatively consistent with the doctor's diagnostic thinking, the training system can use the target medical record as training data to fine-tune the basic network model to obtain a medical record generation model that can better simulate the doctor's diagnostic thinking to determine the patient's problem medical record.

[0089] This embodiment does not limit the type and framework of the basic network model, which can be determined by the training system based on requirements, historical records, experiments, etc.

[0090] Based on the analysis of steps S301 to S304 above, we can see that in this embodiment, the training system corrects the initial medical record based on the medical record template in the standardized medical record database, so that the resulting target medical record is more consistent with the doctor's diagnostic thinking. As a result, the medical record generation model trained based on the target medical record can mimic the doctor's diagnostic thinking and generate and output the corresponding medical record.

[0091] In order to help readers understand the technical principles of the training method provided in this manual more deeply, Figures 4 to 6 Provide a more detailed explanation.

[0092] in, Figure 4 This is a flow chart of a training method for a medical record generation model provided in another embodiment of this specification. Figure 4 As shown, the method includes the following steps S401 to S409:

[0093] S401: Obtain initial patient questions.

[0094] It is understandable that in order to avoid cumbersome descriptions, the technical features that are the same or similar to those in the above examples will not be described in detail in this embodiment.

[0095] For example, regarding the implementation principle of S401, reference may be made to the relevant description of S301 in the above example, which will not be repeated here.

[0096] S402: Collect information based on the initial patient questions to obtain a case feature sample.

[0097] A case characteristic sample can be understood as the core information about a patient's health status extracted through a comprehensive analysis, summary, and organization of the patient's medical history, physical examination results, and auxiliary examination data. For example, a case characteristic sample can include not only the patient's positive findings (i.e., existing symptoms or abnormal findings) but also negative symptoms and signs with differential diagnostic significance (i.e., symptoms or signs that are not present but are important for excluding certain diseases).

[0098] Specifically, a case profile sample can be understood as a structured summary derived from a comprehensive analysis of systematically extracted information, such as medical history, physical examination findings, and the results of various auxiliary examinations. This summary is intended to accurately reflect the patient's condition, helping doctors make an accurate initial diagnosis and providing a basis for further differential diagnosis.

[0099] In short, a case profile can be understood as a detailed and structured description of a patient's condition. By integrating information from multiple sources, it provides a comprehensive understanding of the patient's current health status and points out possible diagnostic directions. This characteristic representation helps improve the efficiency and accuracy of medical decision-making.

[0100] In some embodiments, the training system can determine the initial medical history based on the large model. The functionality of the large model can be encapsulated by an agent, and the large model can be shared by multiple agents.

[0101] like Figure 5 , multiple agents may include an information collection agent.

[0102] Accordingly, the training system can systematically extract information from the patient's questions (i.e., initial patient questions) using the information collection agent to extract details about the patient's medical history, physical examination, and auxiliary examinations. After comprehensive analysis, summarization, and organization, a sample of case characteristics is obtained, including positive findings and negative symptoms and signs with differential diagnostic significance.

[0103] S403: Determine a preliminary diagnosis information sample corresponding to the case feature sample based on a preset knowledge base, wherein the preset knowledge base is used to characterize the correspondence between the disease and the key points of the medical inquiry.

[0104] Key points of a medical interview can be understood as the key information that doctors need to pay attention to and understand when questioning patients. This information is crucial for an accurate diagnosis. For example, key points of a medical interview may include: medical history, past medical history, family medical history, personal lifestyle, physical examination, auxiliary examinations, medication use, etc.

[0105] The preset knowledge base can store various diseases and their corresponding key points for diagnosis. For example, the training system or other system can collect relatively common diseases and their corresponding key points for diagnosis to construct a preset knowledge base.

[0106] The preliminary diagnosis information sample can be understood as the preliminary diagnosis information corresponding to the case feature sample, such as the preliminary diagnosis result and the corresponding diagnosis basis.

[0107] Combining the above examples and Figure 5 ,Multiple agents that share a large model can also include preliminary diagnosis agents.

[0108] Accordingly, the training system may determine preliminary diagnosis information samples based on the preliminary diagnosis agent.

[0109] In some embodiments, S403 may include the following steps 11 to 13:

[0110] Step 11: Determine the suspected preliminary diagnosis information sample corresponding to the case feature sample.

[0111] Combining the above examples and Figure 5 ,The training system can utilize the preliminary diagnosis agent to make a preliminary diagnosis and the corresponding diagnostic basis (i.e., the suspected preliminary diagnosis information sample) for the patient based on the case feature sample.

[0112] Step 12: Obtain the first key points of the medical inquiry corresponding to the suspected preliminary diagnosis information sample from the preset knowledge base.

[0113] Combining the above examples and Figure 5 The training system can use the preliminary diagnosis agent to retrieve the key points of the corresponding disease (ie, the first key points of the diagnosis) from the preset knowledge base according to the preliminary diagnosis result obtained in step 11.

[0114] Step 13: Determine the preliminary diagnostic information sample based on the first consultation points.

[0115] In combination with the above example, the training system can use the preliminary diagnosis agent to hand over the first consultation points to the big model so that the big model can reflect on the suspected preliminary diagnosis information sample based on the first consultation points, such as judging whether the suspected preliminary diagnosis information sample is reasonable.

[0116] If the judgment result is unreasonable, the diagnosis is iterated within the large model until a reasonable preliminary diagnosis information sample is obtained. Conversely, if the judgment result is reasonable, the suspected preliminary diagnosis information sample is determined as the preliminary diagnosis information sample.

[0117] Among them, "reasonable" in this embodiment can be determined by the training system based on needs, historical records, experiments, etc.

[0118] Combined with the above analysis of steps 11 to 13, it can be seen that the training system makes a preliminary diagnosis of the patient and the corresponding diagnostic basis (such as a preliminary diagnosis information sample) based on the characteristics of the case (such as a case feature sample), and retrieves the key points of the corresponding disease from the preset knowledge base based on the results of the preliminary diagnosis (such as the first key point of the consultation), and submits it to the model for reflection. If it is judged to be unreasonable, the diagnosis is iteratively performed within the large model to obtain a reasonable preliminary diagnosis result, so that the final preliminary diagnosis information sample has higher accuracy and reliability.

[0119] S404: Determine the differential diagnosis information sample corresponding to the initial diagnosis information sample based on the preset knowledge graph and the preset knowledge base, wherein the preset knowledge graph is used to characterize the differential diagnosis relationship between different diseases, and the initial medical record includes a case feature sample, a preliminary diagnosis information sample, and a differential diagnosis information sample.

[0120] Differential diagnosis can be understood as making a list of possible diseases by analyzing the patient's symptoms, signs and examination results, and then excluding the diseases that do not meet the conditions one by one to finally determine the most likely diagnosis.

[0121] Accordingly, the preset knowledge graph can represent the relationship between diseases and corresponding differential diagnosis diseases. For example, if a disease is known, other diseases that are similar to it can be identified by querying the preset knowledge graph.

[0122] On this basis, the training system can exclude similar diseases to obtain the final diagnosis result (such as differential diagnosis information sample) corresponding to the initial patient problem.

[0123] Combining the above examples and Figure 5 ,Multiple agents that share a large model can also include differential diagnosis agents.

[0124] Accordingly, the training system can determine differential diagnosis information samples based on the differential diagnosis agent.

[0125] Combined with the analysis of S402 to S404 above, it can be seen that in this embodiment, the training system generates an initial medical record, including case feature samples, preliminary diagnosis information samples, and differential diagnosis information samples, by combining the preset knowledge base and the preset knowledge graph. This allows the initial medical record to demonstrate the doctor's thought process, and this thought process is highly consistent with the doctor's diagnostic thinking, that is, highly consistent with the doctor's logical thinking. As a result, the initial medical record has high accuracy and strong practicality.

[0126] In some embodiments, S404 may include the following steps 21 to 23:

[0127] Step 21: Retrieve a list of diseases to be identified and excluded corresponding to the initial diagnostic information sample from the preset knowledge graph.

[0128] Combining the above examples and Figure 5 ,The training system can utilize the differential diagnosis agent to retrieve the corresponding list of diseases to be identified and excluded from the preset knowledge graph of diseases and corresponding differential diagnosis diseases based on the results of the preliminary diagnosis.

[0129] Step 22: Retrieve the second key points of medical consultation corresponding to each disease in the disease list from the preset knowledge base.

[0130] Correspondingly, the training system can retrieve the corresponding diagnosis points (i.e., the second diagnosis points) for each disease in the disease list from the preset knowledge base based on the disease list obtained in step 21.

[0131] Step 23: Determine a differential diagnosis information sample based on the disease list, the second medical inquiry points, and the differential diagnosis process corresponding to each disease in the disease list.

[0132] In combination with the above example, the training system can use the differential diagnosis agent to hand over the disease list, the second consultation points, and the differential diagnosis process to the big model, so that the big model can reflect on the disease list, the second consultation points, and the differential diagnosis process, such as judging whether the differential diagnosis information sample is reasonable.

[0133] If the judgment result is unreasonable, iterative identification is performed within the large model until it is judged to be reasonable, and then the final differential diagnosis information sample and the corresponding original medical record (i.e., initial medical record) are output to the next step.

[0134] Similarly, "reasonable" in this embodiment can be determined by the training system based on demand, historical records, experiments, etc.

[0135] Combined with the above analysis of steps 21 to 23, it can be seen that the training system retrieves the corresponding list of diseases to be identified and excluded from the knowledge graph of diseases-corresponding differential diagnosis diseases through the results of the preliminary diagnosis (such as the initial diagnostic information sample), and then retrieves the corresponding key points of the medical consultation (such as the second key points of the medical consultation) from the knowledge base based on the disease list, and uses the large model to make a process of identifying and excluding each disease. The disease list, medical consultation points, and differential diagnosis process are submitted to the large model for reflection. If it is judged to be unreasonable, iterative identification is performed within the large model until it is judged to be reasonable. The final diagnosis and the sorted initial medical record are output to the next step. This can make the final differential diagnosis information sample have high accuracy and reliability.

[0136] S405: Determine the difference information between the medical record template and the initial medical record.

[0137] As can be seen from the above example, the medical record template can be determined by the doctor based on his or her diagnostic thinking, while the initial medical record can be determined by simulating the doctor's diagnostic thinking. There may be differences between the medical record template and the initial medical record. In this step, the training system can determine the differences between the medical record template and the initial medical record to obtain information about the differences between the two.

[0138] Continuing with the above example and Figure 5 The agents sharing the large model may also include core agents. The training system can determine difference information based on the core agents. For example, the training system can compare the medical record template with the initial medical record based on the core agents to obtain difference information between the two.

[0139] S406: Determine prompt information according to the difference information.

[0140] S407: Correct the initial medical record according to the prompt information to obtain the target medical record.

[0141] For example, after determining the difference information between the medical record template and the initial medical record, that is, after determining the difference between the diagnostic thinking of the simulated doctor and the diagnostic thinking of the doctor in the real scene, the training system can determine the prompt information based on the difference, and correct the initial medical record obtained through the diagnostic thinking of the simulated doctor based on the prompt information, so that the target medical record is relatively closer to the diagnostic thinking of the doctor in the real scene.

[0142] Combined with the above analysis of S405 to S407, it can be seen that in this embodiment, the training system first compares the differences between the medical records corresponding to the simulated and actual diagnostic approaches, then determines prompt information based on these differences. Finally, using this prompt information, it corrects and adjusts the initial medical record corresponding to the simulated diagnostic approach to obtain a target medical record that is relatively more consistent with the actual diagnostic approach. This ensures that the target medical record is highly consistent with the doctor's diagnostic approach in the real scenario, thereby improving the accuracy and reliability of the target medical record.

[0143] Based on the above analysis, it can be seen that the initial medical record can include a case feature sample, a preliminary diagnosis information sample, and a differential diagnosis information sample. Accordingly, in some embodiments, the medical record template can include a case feature template, a preliminary diagnosis information template, and a differential diagnosis information template. The difference information includes at least one of the following:

[0144] The first difference between the case characteristics template and the case characteristics sample.

[0145] The second difference between the preliminary diagnosis information template and the preliminary diagnosis information sample.

[0146] The third difference between the differential diagnosis information template and the differential diagnosis information sample.

[0147] In combination with the above example, taking the difference information including the first difference as an example:

[0148] The core agent can reflect on the corresponding part of the information collection agent to determine the rationality of the corresponding part of the information collection agent. For example, the core agent can compare the case feature template with the case feature sample, such as obtaining a first difference that represents the difference between the two. If the first difference is large, the core agent can determine that the judgment result is unreasonable and propose improvement suggestions based on the first difference, such as obtaining corresponding prompt information (for ease of distinction, it can be referred to as first prompt information).

[0149] Accordingly, the core agent may feed back the first prompt information to the information collection agent, so that the information collection agent can regenerate a relatively more reasonable case feature sample based on the first prompt information.

[0150] In combination with the above example, taking the case where the difference information includes the second difference as an example:

[0151] The core agent can reflect on the corresponding part of the preliminary diagnosis agent to determine the rationality of the corresponding part of the preliminary diagnosis agent. For example, the core agent can compare the preliminary diagnosis information template with the preliminary diagnosis information sample, such as obtaining a second difference used to characterize the difference between the two. If the second difference is large, the core agent can determine that the judgment result is unreasonable, and the core agent can propose improvement suggestions based on the second difference, such as obtaining corresponding prompt information (for ease of distinction, it can be called second prompt information).

[0152] Accordingly, the core agent may feed back the second prompt information to the preliminary diagnosis agent, so that the preliminary diagnosis agent can regenerate a relatively more reasonable preliminary diagnosis information sample based on the second prompt information.

[0153] In combination with the above example, taking the case where the difference information includes the third difference as an example:

[0154] The core agent can reflect on the corresponding part of the differential diagnosis agent to determine the rationality of the corresponding part of the differential diagnosis agent. For example, the core agent can compare the differential diagnosis information template with the differential diagnosis information sample, such as obtaining a third difference used to characterize the difference between the two. If the third difference is large, the core agent can determine that the judgment result is unreasonable, and the core agent can propose improvement suggestions based on the third difference, such as obtaining corresponding prompt information (for ease of distinction, it can be called third prompt information).

[0155] Accordingly, the core agent may feed back the third prompt information to the differential diagnosis agent, so that the differential diagnosis agent can regenerate a relatively more reasonable differential diagnosis information sample based on the third prompt information.

[0156] That is, after obtaining the initial medical record, the training system retrieves a medical record template corresponding to the final diagnosed disease from the standardized medical record database. The core agent then reflects on the corresponding components of each agent (e.g., the information collection agent, the preliminary diagnosis agent, and the differential diagnosis agent) to compare and judge their rationality. If any are judged to be unreasonable, improvement suggestions are proposed and the corresponding agent is notified to make corrections. This process continues until each agent is judged to be reasonable, ultimately outputting a revised, high-quality medical record (i.e., the target medical record).

[0157] That is, the training system can obtain a target medical record with higher quality by combining one or more of the first difference, the second difference, and the third difference, thereby improving the accuracy, reliability, and effectiveness of the target medical record.

[0158] S408: Obtain quality assessment information of the target medical record.

[0159] Quality assessment information can be understood as information that indicates the quality of the target medical record. For example, quality assessment information can be expressed as a grade, with higher grades indicating relatively higher quality. Alternatively, quality assessment information can be expressed as a score, with higher scores indicating relatively higher quality.

[0160] In some embodiments, the training system may obtain quality assessment information based on a network model or through interaction with a doctor.

[0161] Take the training system to obtain quality assessment information based on the network model as an example:

[0162] The training system may input the target medical record into a preset network model to determine quality assessment information corresponding to the target medical record based on the preset network model.

[0163] Similarly, this embodiment does not limit the architecture and type of the preset network model, which can be determined by the training system based on requirements, historical records, experiments, etc.

[0164] Take the example of training a system to obtain quality assessment information through interaction with doctors:

[0165] The training system can output an interactive interface through a display interface or other means. The interactive interface can include the target medical record and interactive prompt information. Accordingly, the physician can perform corresponding operations in the interactive prompt information based on the target medical record. Accordingly, the training system can determine quality assessment information based on the physician's operations.

[0166] like Figure 6 As shown, after obtaining the initial patient question, the training system can obtain and output the target medical record corresponding to the initial patient question and the corresponding interactive prompt information, such as the helpfulness annotation, through the above method.

[0167] Accordingly, the doctor can select "Yes" or "No" in the interactive prompt to evaluate the quality of the target medical record. For example, if the doctor selects "Yes", it means that the quality of the target medical record is relatively high. Conversely, if the doctor selects "No", it means that the quality of the target medical record is relatively low.

[0168] It is worth noting that the training system can combine the two aforementioned methods to determine quality assessment information. For example, the training system can determine quality assessment information based on the two aforementioned methods using a "simultaneous satisfaction" approach. If the quality assessment information determined by both methods indicates high quality, then the quality assessment information is determined to indicate that the quality of the target medical record is relatively high.

[0169] For another example, the training system can determine quality assessment information based on both of the aforementioned methods using different weighting coefficients for different methods. For example, different weighting coefficients can be assigned to the interaction and network model methods, and the final quality assessment information can be determined based on the respective weighting coefficients and the corresponding quality assessment information.

[0170] Similarly, the weight coefficient can be determined by the training system based on demand, historical records, experiments, etc., and this embodiment does not limit it.

[0171] Based on the above analysis, it can be seen that in this embodiment, the training system determines quality assessment information through network models and / or interactive methods, which can achieve diversity and flexibility in determining quality assessment information. In addition, this interactive method not only continuously improves the reasoning ability of the generation system, but also achieves iterative improvement through human-computer collaboration.

[0172] S409: If the quality assessment information meets the preset quality requirements, the basic network model is fine-tuned according to the target medical record to obtain a medical record generation model.

[0173] Similarly, the preset quality requirements can be determined by the training system based on requirements, historical records, experiments, etc., and this embodiment does not limit this.

[0174] Illustratively, the training system may compare the quality assessment information with preset quality requirements to determine whether the quality assessment information meets the preset quality requirements.

[0175] Correspondingly, the judgment result may be judgment result 1: the quality assessment information meets the preset quality requirements. Then the training system can execute S409, such as using the target medical record as training data to fine-tune the basic network model to obtain a medical record generation model.

[0176] The judgment result may also be judgment result 2: the quality assessment information does not meet the preset quality requirements. In this case, the training system may abandon fine-tuning the basic network model based on the target medical record.

[0177] Based on the above analysis of S408 and S409, it can be seen that in this embodiment, after obtaining the target medical record, the training system can first perform a quality assessment on the target medical record. If the quality of the target medical record meets the preset quality requirements, the training system can fine-tune the basic network model based on the target medical record to obtain a medical record generation model. This can enable the medical record generation model to generate relatively high-quality medical records.

[0178] Furthermore, high-quality medical record generation models can be used to assist doctors in diagnosing medical diseases. By utilizing medical records for medical diagnosis, doctors can more transparently see the diagnostic process of the medical record generation model, thereby increasing the trust in the medical record generation model in assisting medical diagnosis.

[0179] Based on the medical record generation model obtained by training the above training system, this specification also provides a method for generating medical records (hereinafter referred to as the generation method). Regarding the application scenarios of the generation method, please refer to the description of the application scenarios of the training method in the above example, which will not be repeated here.

[0180] In addition, the generation method may be performed by a system for generating medical records (hereinafter referred to as the generation system). The generation system may be the same system as the training system or a different system from the training system. For a description of the generation system, please refer to the above example. The description of the training system will not be repeated here.

[0181] If the generation system is a different system from the training system, in some embodiments, the generation system and the training system may be connected via a communication link. The training system may transmit the generated medical record generation model to the generation system via this communication link. When the generation system needs to generate a medical record, it may generate the medical record using the medical record generation model stored therein.

[0182] In other embodiments, the training system may also provide an invocation interface for invoking the medical record generation model. Accordingly, when the generation system needs to generate a medical record, it can invoke the medical record generation model service through the invocation interface to obtain the corresponding medical record.

[0183] See also Figure 7 , Figure 7 This is a flow chart of the method for generating a medical record provided in the embodiment of this specification. Figure 7 As shown, the method includes:

[0184] S701: Obtain target patient questions.

[0185] Similarly, for understanding S701, please refer to the relevant description of obtaining the initial patient problem in S301 in the above example, which will not be repeated here.

[0186] S702: Input the target patient's problem into the medical record generation model to simulate the doctor's diagnostic thinking based on the medical record generation model, and generate and output the medical record corresponding to the target patient's problem.

[0187] The medical record generation model is trained based on the training method described in any of the above embodiments.

[0188] For example, combined Figure 8 It can be seen that the input of the medical record generation model is the target patient problem, and the output is the corresponding medical record (or called artificial intelligence (AI) diagnostic reasoning).

[0189] In some embodiments, the generation system can also output a page including AI-assisted diagnosis information based on the corresponding medical records. Figure 8 As shown in Figure 2, AI-assisted diagnosis information includes the results of preliminary diagnosis and differential diagnosis. Figure 8 As shown, the preliminary diagnosis may include primary / secondary diagnoses.

[0190] Correspondingly, such as Figure 8 As shown in the figure, doctors can perform a "delete" operation on a page containing AI-assisted diagnosis information. Doctors can also search for corresponding standard diagnoses by entering the corresponding diagnosis name, keyword, abbreviation, etc. in the search box to assist doctors in diagnosis, thereby promoting the widespread application of medical assistance systems (such as generation systems) in real hospitals.

[0191] As can be seen from the above examples, the medical record generation model can closely simulate a doctor's diagnostic thinking. Therefore, by using the target patient's problem as input to the medical record generation model, the generation system can produce medical records that closely align with the doctor's diagnostic thinking in the displayed scenario. This improves the accuracy and reliability of medical records, and furthermore, achieves a significant breakthrough in the accuracy and practicality of the generation system.

[0192] In other words, the generation system, based on the patient record generation model, not only emulates the physician's thought process but also generates a chain of diagnostic evidence and detailed reasoning paths, enabling physicians to easily understand and participate in the repair and optimization process. Furthermore, it incorporates the complex logical relationships underlying medical knowledge and the physician's personal knowledge, making the generated results more transparent, accurate, and easily accepted by clinicians. Furthermore, the generation system can interact with physicians based on the output of the patient record generation model, allowing them to review and correct the results, ensuring the credibility and clinical value of the content.

[0193] In this specification, the large model can be an abbreviation of the large language model (LLM). The large language model is a natural language processing model based on deep learning technology. Its parameter magnitude usually reaches billions to hundreds of billions or even higher, and it has powerful language understanding and generation capabilities. The large language model can adopt the Transformer architecture or its variants (such as GPT, BERT, etc.). The architecture uses the attention mechanism to achieve global modeling of sequence data, which can efficiently handle long-distance dependencies, thereby performing well in natural language tasks. The large language model learns the statistical characteristics and semantic relevance of the language by pre-training on a large-scale corpus, giving it excellent generalization capabilities. The core capabilities of the large language model include but are not limited to: understanding contextual semantics, generating coherent and grammatically correct text, performing logical reasoning, and handling multi-task scenarios. Its usage methods generally include two modes: direct inference and fine-tuning. In direct inference mode, the user guides the large language model to generate specific outputs by designing prompts. The prompts can be textual task descriptions or instructions, stimulating the semantic understanding and generation capabilities of the large language model. In fine-tuning mode, the large language model is further trained on a small-scale dataset in a specific domain to optimize its performance on a specific task. The powerful generalization and flexibility of large language models make them a valuable tool in the field of artificial intelligence technology, providing efficient and accurate solutions for automated text generation and comprehension.

[0194] In some embodiments, the large language model may also have the ability to understand and generate data from other modalities (such as vision, audio, etc.). In this case, the large language model may also be called a multimodal large language model (MLLMs). MLLMs provide a richer and more natural interactive experience by integrating multiple types of inputs and outputs such as text, images, and sounds. The core advantage of MLLMs is that they can process and understand information from different modalities and fuse this information to complete complex tasks. For example, MLLMs can analyze a picture and generate descriptive text, or generate corresponding images based on the text description. This cross-modal understanding and generation capability gives MLLMs broad application prospects in multiple fields.

[0195] It should be noted that the key technologies of large language models can be found in the detailed description in the paper "A Survey of Large Language Models" (paper number: arXiv:2303.18223v16, published on March 11, 2025, public link: https: / / doi.org / 10.48550 / arXiv.2303.18223), which will not be repeated in this manual.

[0196] It is worth noting that the above examples are only used to illustrate possible implementations of the training method and generation method of this specification, and should not be understood as limiting the implementations of the training method and generation method of this specification. For example, based on the above technical concept, some of the above technical features can be combined to obtain a new embodiment; new technical features can be added to the above examples to obtain a new embodiment; some technical features can be reduced to obtain a new embodiment based on the above examples; some technical features in the above examples can be replaced with other technical features; some technical features and their order in the above examples can be adjusted to obtain a new embodiment, etc., which will not be listed here one by one.

[0197] Based on the above technical concept, this specification also provides a computer-readable non-temporary storage medium, which stores at least one instruction set. When the at least one instruction set is executed by the processor, the steps of the training method and generation method described in this specification are implemented.

[0198] In some possible implementations, various aspects of this specification may also be implemented in the form of a program product comprising program code. Taking training system 200 as an example, when the program product is executed on training system 200, the program code is used to cause training system 200 to perform the steps of the training method described in this specification. The program product for implementing the above method may utilize a portable compact disc read-only memory (CD-ROM) comprising the program code and may be executed on training system 200. However, the program product of this specification is not limited thereto. In this specification, a readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system. The program product may utilize any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or any combination thereof. More specific examples of computer-readable storage media include: an electrical connection having one or more conductors, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. The computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, which carries readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The computer-readable storage medium may also be any readable medium other than a readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical cable, RF, etc., or any suitable combination of the foregoing. Program code for performing the operations described herein may be written in any combination of one or more programming languages, including object-oriented programming languages such as Java, C++, and conventional procedural programming languages such as C or similar programming languages. The program code may be executed entirely on the training system 200, partially on the training system 200, as a stand-alone software package, partially on the training system 200 and partially on a remote training system, or entirely on the remote training system 200.

[0199] In addition, for the description of the generation system, please refer to the above example and will not be repeated here.

[0200] It should be noted that the collection, storage, use, processing, transmission, provision and disclosure of relevant user information (such as questions initiated by patients, etc.) involved in the technical solution of this specification are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0201] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the particular order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0202] In summary, after reading this detailed disclosure, those skilled in the art will appreciate that the foregoing detailed disclosure may be presented by way of example only and may not be limiting. Although not expressly stated herein, those skilled in the art will understand that this specification encompasses various reasonable changes, improvements, and modifications to the embodiments. Such changes, improvements, and modifications are intended to be suggested by this specification and are within the spirit and scope of the exemplary embodiments of this specification.

[0203] Furthermore, certain terms in this specification have been used to describe embodiments of this specification. For example, “one embodiment,” “an embodiment,” and / or “some embodiments” mean that a particular feature, structure, or characteristic described in connection with that embodiment may be included in at least one embodiment of this specification. Therefore, it is emphasized and should be understood that two or more references to “an embodiment,” “one embodiment,” or “an alternative embodiment” in various parts of this specification do not necessarily refer to the same embodiment. Furthermore, particular features, structures, or characteristics may be appropriately combined in one or more embodiments of this specification.

[0204] It should be understood that in the foregoing descriptions of the embodiments of this specification, to facilitate understanding of a feature and to simplify this specification, various features are combined in a single embodiment, figure, or description thereof. However, this does not necessarily mean that these features are combined. When reading this specification, a person skilled in the art may label some of the devices as separate embodiments. In other words, the embodiments of this specification can also be understood as the integration of multiple sub-embodiments. The content of each sub-embodiment is also valid even when it includes fewer than all the features of a single previously disclosed embodiment.

[0205] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, and the like, cited herein (excluding any historical review documents related thereto) is hereby incorporated by reference for all purposes relevant to this document, such as within the specification and claims herein. However, if there is any inconsistency or conflict between the descriptions, definitions, and / or terminology of such materials and the descriptions, definitions, and / or terminology used herein, the descriptions, definitions, and / or terminology used herein shall control.

[0206] Finally, it should be understood that the embodiments of the application disclosed herein are illustrative of the principles of the embodiments of this specification. Other modified embodiments are also within the scope of this specification. Therefore, the embodiments disclosed in this specification are merely examples and not limitations. Those skilled in the art can adopt alternative configurations based on the embodiments in this specification to implement the application in this specification. Therefore, the embodiments of this specification are not limited to the embodiments precisely described in the application.

Claims

1. A training method for a medical record generation model, comprising: Obtaining an initial patient problem and obtaining an initial medical record corresponding to the initial patient problem determined by simulating a doctor's diagnostic thinking, wherein the initial medical record is used to represent a diagnostic evidence chain corresponding to the diagnostic thinking; Retrieving a medical record template corresponding to the initial medical record from a pre-constructed standardized medical record database, wherein the standardized medical record database includes a template corresponding to each medical record determined based on the doctor's diagnostic thinking; Correcting the initial medical record according to the medical record template to obtain a target medical record; and The basic network model is fine-tuned according to the target medical record to obtain a medical record generation model, wherein the medical record generation model is used to simulate the doctor's diagnostic thinking and generate and output the medical record corresponding to the target patient's problem.

2. The method according to claim 1, wherein The obtaining of the initial medical record corresponding to the initial patient problem determined by the diagnostic thinking of the simulated doctor includes: Collect information based on the initial patient questions to obtain a sample of case characteristics; Determining a preliminary diagnostic information sample corresponding to the case feature sample based on a preset knowledge base, wherein the preset knowledge base is used to characterize the correspondence between diseases and key points of medical consultation; and According to the preset knowledge graph and the preset knowledge base, a differential diagnosis information sample corresponding to the initial diagnosis information sample is determined, wherein the preset knowledge graph is used to characterize the differential diagnosis relationship between different diseases, and the initial medical record includes the case feature sample, the preliminary diagnosis information sample, and the differential diagnosis information sample.

3. The method according to claim 2, wherein: Correcting the initial medical record according to the medical record template to obtain a target medical record includes: Determining difference information between the medical record template and the initial medical record; Determining prompt information according to the difference information; and The initial medical record is corrected according to the prompt information to obtain the target medical record.

4. The method according to claim 3, wherein: The medical record template includes a case feature template, a preliminary diagnosis information template, and a differential diagnosis information template; the difference information includes at least one of the following: a first difference between the case characteristic template and the case characteristic sample; a second difference between the preliminary diagnostic information template and the preliminary diagnostic information sample; A third difference between the differential diagnosis information template and the differential diagnosis information sample.

5. The method according to claim 2, wherein: The determining of the preliminary diagnostic information sample corresponding to the case feature sample based on the preset knowledge base includes: Determine a sample of suspected preliminary diagnosis information corresponding to the sample of case characteristics; Obtaining a first medical inquiry point corresponding to the suspected preliminary diagnosis information sample from the preset knowledge base; and The preliminary diagnosis information sample is determined according to the first medical inquiry points.

6. The method according to claim 2, wherein: The determining, based on the preset knowledge graph and the preset knowledge base, a differential diagnosis information sample corresponding to the initial diagnosis information sample includes: Retrieving from the preset knowledge graph a list of diseases to be identified and excluded corresponding to the initial diagnostic information sample; Retrieving from the preset knowledge base the second key points of medical consultation corresponding to each disease in the disease list; and The differential diagnosis information sample is determined according to the disease list, the second medical inquiry points, and the differential diagnosis process corresponding to each disease in the disease list.

7. The method according to any one of claims 1 to 6, wherein The method further comprises: Obtaining quality assessment information of the target medical record; And, the fine-tuning of the basic network model according to the target medical record to obtain the medical record generation model includes: if the quality assessment information meets the preset quality requirements, fine-tuning the basic network model according to the target medical record to obtain the medical record generation model.

8. The method according to claim 7, wherein: The obtaining of quality assessment information of the target medical record includes at least one of the following: Obtaining the quality assessment information through a preset network model; The quality assessment information is obtained based on interaction with a physician.

9. The method according to any one of claims 1 to 6, wherein Each template in the standardized medical record database is generated by doctors by marking corresponding diseases based on diagnostic ideas.

10. A method for generating a medical record, comprising: Obtain targeted patient questions; Inputting the target patient's problem into a medical record generation model, simulating the doctor's diagnostic thinking based on the medical record generation model, and generating and outputting a medical record corresponding to the target patient's problem; The medical record generation model is obtained by training based on the training method according to any one of claims 1 to 9.

11. A training system for a medical record generation model, comprising: At least one storage medium storing at least one instruction set for training a medical record generation model; At least one processor is communicatively connected to the at least one storage medium, wherein when the at least one processor is running, it reads the at least one instruction set and executes the training method as described in any one of claims 1 to 9 according to the instructions of the at least one instruction set.

12. A system for generating medical records, comprising: At least one storage medium storing at least one instruction set for training a medical record generation model; At least one processor is communicatively connected to the at least one storage medium, wherein when the at least one processor is running, it reads the at least one instruction set and executes the generation method according to the instructions of the at least one instruction set.

Images