Disease assessment system and method, electronic equipment and medium

Abstract

The invention provides a disease assessment system and method, an electronic device and a medium wherein a data acquisition module is used for acquiring form data and clinical text record data of a target patient; the large model reasoning module is connected with the data acquisition module, and the large model reasoning module is used for performing semantic understanding and reasoning on the clinical text record data based on the medical knowledge base and outputting at least one intermediate semantic understanding result; the rule engine module is connected with the data acquisition module and the large model reasoning module, and the rule engine module is used for receiving the form data and the at least one intermediate semantic understanding result and performing combinational logic operation on the at least one intermediate semantic understanding result according to a predefined evaluation rule; and the assessment result generation module is used for generating a final disease risk assessment conclusion of the target patient according to the logical operation result of the rule engine module. According to the invention, automatic and accurate disease risk assessment is realized, and the clinical decision-making efficiency and quality are improved.

Description

Technical Field

[0001] This application relates to the field of disease assessment technology, and more specifically, to a disease assessment system, method, electronic device, and medium. Background Technology

[0002] Venous thromboembolism (VTE) is the third leading cause of cardiovascular death worldwide, often referred to as a "silent killer" in hospitals due to its insidious onset and rapid progression. Currently, clinical practice primarily relies on standardized scoring scales such as Caprini and Padua for VTE risk assessment. This process, entirely dependent on healthcare professionals manually collecting patient information and calculating scores, has significant limitations: firstly, it is inefficient, with manual operations being time-consuming and labor-intensive, resulting in generally low assessment rates; secondly, it relies on subjective judgment, easily leading to missed or incorrect assessments; and thirdly, it struggles to utilize unstructured data, as the assessments primarily rely on structured data, failing to effectively mine and utilize the non-textual content of electronic medical records, such as progress notes and imaging reports, which contain a wealth of crucial information. Summary of the Invention

[0003] In view of the above, the purpose of this application is to provide a disease assessment system, method, electronic device and medium that aims to overcome at least one of the above-mentioned defects.

[0004] In a first aspect, this application provides a disease assessment system, which includes a data acquisition module, a large model inference module, a rule engine module, and an assessment result generation module. The data acquisition module is used to acquire form data and clinical text record data of the target patient. The large model reasoning module is connected to the data acquisition module. The large model reasoning module is used to perform semantic understanding and reasoning on the clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result. The rule engine module is connected to the data acquisition module and the large model inference module. The rule engine module is used to receive the form data and the at least one intermediate semantic understanding result, and to perform combinational logic operations on the at least one intermediate semantic understanding result according to predefined evaluation rules. The assessment result generation module is connected to the rule engine module, and the assessment result generation module is used to generate the final disease risk assessment conclusion of the target patient based on the logical operation results of the rule engine module.

[0005] In one possible implementation, the system further includes: An assessment management module, which is connected to the rule engine module, is used to send assessment instructions to the rule engine module at preset time nodes, such as when the target patient is admitted to the hospital, after surgery, or before discharge, to trigger the disease risk assessment process for the target patient.

[0006] In one possible implementation, the large model inference module includes: The prompt word construction unit is used to combine the clinical text record data, the predefined assessment task description and the relevant medical knowledge provided by the medical knowledge base to generate prompt word instructions; The model interaction unit is used to send the prompt word instruction to the local large language model and receive the response text containing reasoning analysis returned by the local large language model to generate an intermediate semantic understanding result.

[0007] In one possible implementation, the model interaction unit processes the response text of the large language model in the following manner: The local large language model is required to return a response in a predefined JSON format; The returned JSON string is format-validated and parsed to extract the Boolean fields representing the judgment conclusion and the text fields representing the reasoning basis.

[0008] In one possible implementation, the rule engine module performs logical operations in the following manner: Multiple judgment logic units are configured, and each judgment logic unit is used to call the large model inference module once to obtain an intermediate semantic understanding result for a specific clinical problem. Use logical connectors to combine multiple intermediate semantic understanding results.

[0009] In one possible implementation, the system further includes: The traceability quality control module is connected to the data acquisition module, the large model inference module, the rule engine module, and the evaluation result generation module. It is used to record and associate the data throughout the entire process from data input, model inference, rule operation to conclusion generation, so as to build a traceable evaluation evidence chain.

[0010] In one possible implementation, the assessment rules are constructed based on clinical treatment guidelines and risk assessment scales for the target disease.

[0011] Secondly, this application provides a disease assessment method, the method comprising: The data acquisition module is used to acquire form data and clinical text record data of the target patients; The large model reasoning module is connected to the data acquisition module. The large model reasoning module is used to perform semantic understanding and reasoning on the clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result. The rule engine module is connected to the data acquisition module and the large model inference module. The rule engine module is used to receive the form data and the at least one intermediate semantic understanding result, and to perform combinational logic operations on the at least one intermediate semantic understanding result according to predefined evaluation rules. The assessment result generation module is connected to the rule engine module, and the assessment result generation module is used to generate the final disease risk assessment conclusion of the target patient based on the logical operation results of the rule engine module.

[0012] Thirdly, this application also provides an electronic device, including: a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory via the bus, and when the machine-readable instructions are executed by the processor, the steps of the method described above are performed.

[0013] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the method described above.

[0014] This application provides a disease assessment system, method, electronic device, and medium. A data acquisition module acquires form data and clinical text record data from the target patient. A large-scale model inference module, connected to the data acquisition module, performs semantic understanding and inference on the clinical text record data based on a medical knowledge base, outputting at least one intermediate semantic understanding result. A rule engine module, connected to the data acquisition module and the large-scale model inference module, receives form data and at least one intermediate semantic understanding result, and performs combinational logic operations on the at least one intermediate semantic understanding result according to predefined assessment rules. An assessment result generation module generates a final disease risk assessment conclusion for the target patient based on the logic operation results of the rule engine module. This application achieves automatic and accurate disease risk assessment, improving the efficiency and quality of clinical decision-making.

[0015] 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

[0016] 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.

[0017] Figure 1 This is a schematic diagram of the structure of a disease assessment system provided in an embodiment of this application; Figure 2 A flowchart illustrating a disease assessment method provided in this application embodiment; Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. Based on the embodiments of this application, every other embodiment obtained by those skilled in the art without inventive effort falls within the scope of protection of this application.

[0019] First, the applicable scenarios for this application will be introduced. This application can be applied to the field of disease assessment technology.

[0020] Research has revealed that venous thromboembolism (VTE) is the third leading cause of cardiovascular death worldwide. Due to its insidious onset, rapid progression, and high mortality risk, it is often referred to as the "silent killer" in hospitals. Statistics show that up to 60% of VTE cases occur during hospitalization or after discharge, posing a significant threat to medical quality and safety. Therefore, the National Health Commission has included "improving the standardized prevention rate of venous thromboembolism" in the national medical quality and safety improvement goals for three consecutive years, explicitly requiring medical institutions to strengthen VTE prevention and control through information technology.

[0021] In traditional medical practice, VTE risk assessment mainly relies on standardized scoring systems such as Caprini, Padua, and Khorana. Healthcare professionals manually collect risk factors such as patient age, surgical history, cancer history, and mobility, and then manually score them. However, this manual assessment model has significant limitations: the process is time-consuming and labor-intensive, the assessment rate is generally low (only 53.75% in some hospitals), and it is easily affected by subjective factors, leading to missed or incorrect assessments.

[0022] Existing scoring systems primarily rely on structured data, making it difficult to effectively utilize the vast amounts of unstructured text data in electronic medical records, such as medical progress notes and imaging reports, which contain a wealth of crucial clinical information. Furthermore, manual assessment is inefficient, time-consuming, and ill-suited to the differentiated assessment needs of various specialties (such as oncology, obstetrics and gynecology, and orthopedics), especially for specific populations like cancer patients and pregnant women, where flexible, personalized assessment and treatment strategies are lacking.

[0023] Based on this, embodiments of this application provide a disease assessment system, method, electronic device, and medium, which aim to achieve in-depth analysis and intelligent assessment of multi-source medical data through the semantic understanding and reasoning capabilities of a large language model, providing a more accurate and efficient solution for VTE prevention and treatment.

[0024] Please see Figure 1 , Figure 1 This is a schematic diagram of the structure of a disease assessment system provided in an embodiment of this application. Figure 1 As shown in the figure, the disease assessment system provided in this application embodiment includes: a data acquisition module 101, a large model inference module 102, a rule engine module 103, and an assessment result generation module 104.

[0025] The data acquisition module 101 is used to acquire form data and clinical text record data of the target patient.

[0026] The data acquisition module 101 serves as the foundational service for the system to acquire patient medical record data. It is responsible for connecting with various hospital information systems to automate data collection and standardized cleaning. Specifically, it acquires form data (such as basic patient information, vital signs, and laboratory test results) and clinical text record data (such as admission records, progress notes, and imaging reports) from target patients. The data acquisition module 101 primarily achieves secure connection with data sources such as the Hospital Information System (HIS) and Electronic Medical Records (EMR), automatically collecting various clinical data from admitted patients. For the collected raw data, the data acquisition module 101 performs a series of preprocessing operations, including field transformation and mapping, data cleaning (handling missing and outlier values), and finally persistently stores the processed data. This process ensures data quality and consistency, providing high-quality data input for subsequent intelligent assessment.

[0027] The large model reasoning module 102 is connected to the data acquisition module 101. The large model reasoning module 102 is used to perform semantic understanding and reasoning on clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result.

[0028] In a preferred example of this application, the large model reasoning module 102 includes a prompt word construction unit and a model interaction unit.

[0029] The prompt word construction unit combines clinical text record data, predefined assessment task descriptions, and relevant medical knowledge provided by the medical knowledge base to generate prompt word instructions. The model interaction unit sends the prompt word instructions to the local large language model and receives the response text containing reasoning analysis returned by the local large language model to generate an intermediate semantic understanding result.

[0030] Specifically, the large-scale model reasoning module 102 dynamically integrates patient data, assessment tasks, and the medical knowledge base through a prompt word construction unit to generate precise instructions. Then, the model interaction unit calls the large model, parsing its returned natural language response into structured (Boolean values ​​+ supporting evidence) intermediate results for use by the rule engine. The medical knowledge base automatically extracts and structures core knowledge such as disease diagnostic criteria, typical clinical manifestations, and key risk assessment elements by inputting authoritative treatment guidelines, clinical pathways, and other documents into the large model, transforming them into a dynamic knowledge base that is machine-understandable and applicable.

[0031] In a preferred example of this application, the model interaction unit processes the response text of the large language model in the following manner: The local large language model is required to return a response in a predefined JSON format; the returned JSON string is then format-validated and parsed to extract the Boolean fields representing the judgment conclusion and the text fields representing the reasoning basis.

[0032] Among them, the model interaction unit uses a format conversion tool to strictly verify the structure of the returned data and parses the JSON attributes according to the predetermined format, converting them into executable objects, thereby adding a programmable and automated quality control barrier to the AI ​​evaluation results.

[0033] Specifically, the large model inference module 102 receives requests from the rule engine module 103, is responsible for constructing accurate prompt words, and manages communication with the large model service. The prompt word template integrates patient data, assessment task descriptions, and knowledge base content, and explicitly requires the large model to return results in a specified JSON format. This is necessary for subsequent parsing. This module includes successful validation of string-to-JSON conversion and sets the number of failed attempts to accommodate the uncertainty of the probability statistics of large model inference. The large model inference module 102 adopts streaming requests to cope with the latency of large model response, ensuring stable overall system throughput and response speed in high-concurrency scenarios, thereby supporting frequent calls to the rule engine.

[0034] The rule engine module 103 is connected to the data acquisition module 101 and the large model reasoning module 102. The rule engine module 103 is used to receive form data and at least one intermediate semantic understanding result, and to perform combinational logic operations on at least one intermediate semantic understanding result according to predefined evaluation rules.

[0035] Specifically, the rule engine module 103 encapsulates all the complex details of interacting with the large model through judgment logic units (or custom functions). These judgment logic units define core parameters including: patient hospitalization serial number, judgment method prompts, and disease knowledge base context. The function return value is conventionally a Boolean value. The rule engine supports AND, OR, and NOT logical connectors, enabling the connection of outputs from multiple judgment logic units to handle complex clinical judgment conditions. The evaluation rules are constructed based on clinical treatment guidelines and risk assessment scales for the target disease. Specifically, the evaluation rules directly transform the medical standards in clinical treatment guidelines and risk assessment scales (such as the Caprini scoring system) into executable "if-then" logical conditions, ensuring the authority and standardization of the evaluation.

[0036] In a preferred embodiment of this application, the rule engine module 103 performs logical operations in the following manner: Multiple decision logic units are configured, each of which calls the large model inference module 102 once to obtain an intermediate semantic understanding result for a specific clinical problem; multiple intermediate semantic understanding results are combined using logical connectors.

[0037] Here, a specific clinical problem refers to the smallest, independently identifiable medical question extracted from the complete clinical assessment scale. For example, in the assessment, "Is the patient obese?" or "Does the patient have a lower limb cast?". Each decision logic unit is specifically responsible for posing such an atomic question to the larger model and returning its "yes / no" answer as a Boolean value, which the rule engine then combines to complete the final determination of complex assessment items (such as the Caprini score).

[0038] Specifically, the rule engine module 103 combines the Aviator rule engine (a rule engine middleware in Java) with the capabilities of the large model, enabling flexible and efficient execution of the evaluation logic. It encapsulates all the complex details of interaction with the large model through a judgment logic unit, i.e., a custom function. This custom function defines three core parameters: (1) Patient hospitalization serial number: used to uniquely identify and obtain preprocessed patient data. (2) Judgment method prompt: This is a configurable instruction used to guide the focus and direction of the large model's current evaluation, such as "determine whether the patient is at risk of stroke." (3) Disease knowledge base context: Relevant clinical manifestations, diagnostic criteria, and other knowledge extracted from the medical knowledge base serve as a reference for the large model's analysis and reasoning.

[0039] The function returns a Boolean value, and the final result is sent to the rule engine. In particular, the rule engine supports AND, OR, and NOT logical connectors, which can connect multiple logical units to obtain the result in complex situations where it is difficult to make a single judgment. For example, one item in the Caprini (surgical) scoring table of VTE: "Multiple traumatic fractures of the hip, pelvis, or lower extremities".

[0040] The assessment result generation module 104 is connected to the rule engine module 103. The assessment result generation module 104 is used to generate the final disease risk assessment conclusion for the target patient based on the logical operation results of the rule engine module 103. The final disease risk assessment conclusion refers to the probability that the patient will eventually develop the disease.

[0041] In a preferred embodiment of this application, the system further includes an assessment management module, which is connected to the rule engine module 103 and is used to send assessment instructions to the rule engine module 103 at preset time nodes, such as when the target patient is admitted to the hospital, after surgery, or before discharge, to trigger the disease risk assessment process for the target patient.

[0042] The traceability quality control module is connected to the data acquisition module 101, the large model reasoning module 102, the rule engine module 103, and the evaluation result generation module 104. It is used to record and associate the data throughout the entire process from data input, model reasoning, rule operation to conclusion generation, in order to build a traceable evaluation evidence chain.

[0043] Specifically, the assessment management module and the traceability quality control module are responsible for managing the entire assessment lifecycle and ensuring the traceability of the assessment process. The assessment management module is mainly responsible for triggering and controlling the assessment process. It can automatically trigger the initial assessment, key node reassessment, and final discharge node quality control assessment results based on important nodes in the patient's medical treatment, such as the initial assessment upon admission, the secondary assessment after surgery, and the final assessment before discharge.

[0044] In terms of quality control and traceability, the traceability and quality control module records a complete record of each assessment, including the input patient data, the prompts sent to the large model, the original return results from the large model, and the final assessment conclusion. This provides a complete chain of evidence for the assessment results, facilitating review and auditing by medical staff and meeting the stringent requirements of medical quality management.

[0045] As an example, the functions and collaborative relationships of each module will be explained in detail below, taking into account the specific workflow of the system.

[0046] First, this application selects and deploys open-source large-scale language models. Local large-scale language models are preferably Deepseek-R1, DeepSeek-V3, or Qwen3, and are placed on the hospital's local server or in a private cloud environment to ensure the security of patient data and low processing latency.

[0047] After the system is ready, the data acquisition module 101 enters the crucial data preparation and processing stage. This stage requires collecting all of the patient's medical data from various hospital information systems. This data is carefully organized and converted into a structured JSON format. The core data object is patientBase, which is like a digital portfolio containing all the patient's information. The outermost layer of this portfolio contains the patient's basic hospitalization information, while its interior contains several important sub-documents: first, the medDoc sub-document, which contains the main text of the medical record; second, the surgery sub-document, which records all surgical information; third, the testReport sub-document, which encompasses all laboratory report results; and finally, the order sub-document, which details the treatment orders.

[0048] At the same time, the rule engine module 103 needs to prepare another JSON object named ruleRequest, which represents a specific task that the system is required to complete. This object clearly indicates what disease item needs to be evaluated, what the specific requirements for the item are, and together with the relevant medical knowledge background provided by the medical knowledge base, it constitutes a complete definition of the evaluation task.

[0049] Next, the inference request sent to the large model is constructed, i.e., the prompt word is assembled. This step is completed by the large model inference module 102 in conjunction with the rule engine module 103. This application uses clear Markdown syntax to construct the prompt word because the large model understands paragraph headings marked with hash symbols better. The prompt word explicitly requires the large model to perform a disease inference task, and details the structure and meaning of the data source. Then, the complete character content of the two JSON objects, patientBase and ruleRequest, is directly concatenated. Finally, strict requirements are placed on the output results, including the requirement to return standard JSON format, the definition of each field (such as item, confirm, reason, documentOrigin), and special emphasis is placed on the fact that the output JSON must be wrapped in code markup to ensure format integrity.

[0050] Then, the system enters the core reasoning phase. The large-scale model reasoning module 102 submits this carefully assembled prompt to the deployed large-scale model for reasoning analysis. Based on the provided needs, medical knowledge, and detailed medical record information, the large-scale model performs comprehensive logical analysis and reasoning, ultimately forming its conclusion on whether the patient has the disease.

[0051] Finally, the system enters the result processing and integration phase. The large model inference module 102 receives and rigorously verifies the JSON results returned by the large model, parsing them into structured intermediate semantic understanding results. Subsequently, the rule engine module 103 receives this intermediate result and performs combinational logic operations using its Boolean values. Meanwhile, the evaluation management module and the traceability quality control module completely record and store all data from the entire evaluation process, including inputs, prompts, the model's original response, and the final conclusion, forming a traceable chain of evidence. At this point, a complete automated evaluation process concludes, and the system successfully integrates the AI's inference results into the clinical workflow, providing intelligent decision support for medical staff.

[0052] Based on the same inventive concept, this application also provides a disease assessment method corresponding to the disease assessment system. Since the principle of the device in this application is similar to that of the disease assessment system described above, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be described again.

[0053] Please see Figure 2 , Figure 2 A flowchart of a disease assessment method provided in this application embodiment is shown below. Figure 2 As shown, the disease assessment methods include: S101, the data acquisition module is used to acquire form data and clinical text record data of the target patient.

[0054] S102. The large model reasoning module is connected to the data acquisition module. The large model reasoning module is used to perform semantic understanding and reasoning on clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result.

[0055] S103, the rule engine module is connected to the data acquisition module and the large model reasoning module. The rule engine module is used to receive form data and at least one intermediate semantic understanding result, and to perform combinational logic operations on at least one intermediate semantic understanding result according to predefined evaluation rules.

[0056] S104. The assessment result generation module is connected to the rule engine module. The assessment result generation module is used to generate the final disease risk assessment conclusion for the target patient based on the logical operation results of the rule engine module.

[0057] Please see Figure 3 , Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 3 As shown, the electronic device 300 includes a processor 310, a memory 320, and a bus 330.

[0058] The memory 320 stores machine-readable instructions that can be executed by the processor 310. When the electronic device 300 is running, the processor 310 and the memory 320 communicate via the bus 330. When the machine-readable instructions are executed by the processor 310, the steps of the method described above can be performed. For specific implementation details, please refer to the method embodiment, which will not be repeated here.

[0059] This application also provides a computer-readable storage medium storing a computer program. When the computer program is run by a processor, it can execute the steps of the method described above. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0060] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0061] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the shown or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0062] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0063] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0064] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0065] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The scope of protection of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application. Such modifications, changes, 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 this application, and should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A disease assessment system, characterized in that, The system includes a data acquisition module, a large model inference module, a rule engine module, and an evaluation result generation module. The data acquisition module is used to acquire form data and clinical text record data of the target patient; The large model reasoning module is connected to the data acquisition module. The large model reasoning module is used to perform semantic understanding and reasoning on the clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result. The rule engine module is connected to the data acquisition module and the large model inference module. The rule engine module is used to receive the form data and the at least one intermediate semantic understanding result, and to perform combinational logic operations on the at least one intermediate semantic understanding result according to predefined evaluation rules. The assessment result generation module is connected to the rule engine module, and the assessment result generation module is used to generate the final disease risk assessment conclusion of the target patient based on the logical operation results of the rule engine module.

2. The system according to claim 1, characterized in that, The system also includes: An assessment management module, which is connected to the rule engine module, is used to send assessment instructions to the rule engine module at preset time nodes, such as when the target patient is admitted to the hospital, after surgery, or before discharge, to trigger the disease risk assessment process for the target patient.

3. The system according to claim 1, characterized in that, The large model inference module includes: The prompt word construction unit is used to combine the clinical text record data, the predefined assessment task description and the relevant medical knowledge provided by the medical knowledge base to generate prompt word instructions; The model interaction unit is used to send the prompt word instruction to the local large language model and receive the response text containing reasoning analysis returned by the local large language model to generate an intermediate semantic understanding result.

4. The system according to claim 3, characterized in that, The model interaction unit processes the response text of the large language model in the following manner: The local large language model is required to return a response in a predefined JSON format; The returned JSON string is format-validated and parsed to extract the Boolean fields representing the judgment conclusion and the text fields representing the reasoning basis.

5. The system according to claim 1, characterized in that, The rule engine module performs logical operations in the following ways: Multiple judgment logic units are configured, and each judgment logic unit is used to call the large model inference module once to obtain an intermediate semantic understanding result for a specific clinical problem. Use logical connectors to combine multiple intermediate semantic understanding results.

6. The system according to claim 1, characterized in that, The system also includes: The traceability quality control module is connected to the data acquisition module, the large model inference module, the rule engine module, and the evaluation result generation module. It is used to record and associate the data throughout the entire process from data input, model inference, rule operation to conclusion generation, so as to build a traceable evaluation evidence chain.

7. The system according to claim 1, characterized in that, The assessment rules are constructed based on clinical treatment guidelines and risk assessment scales for the target disease.

8. A disease assessment method, characterized in that, The method includes: The data acquisition module is used to acquire form data and clinical text record data of the target patients; The large model reasoning module is connected to the data acquisition module. The large model reasoning module is used to perform semantic understanding and reasoning on the clinical text record data based on the medical knowledge base, and output at least one intermediate semantic understanding result. The rule engine module is connected to the data acquisition module and the large model inference module. The rule engine module is used to receive the form data and the at least one intermediate semantic understanding result, and to perform combinational logic operations on the at least one intermediate semantic understanding result according to predefined evaluation rules. The assessment result generation module is connected to the rule engine module, and the assessment result generation module is used to generate the final disease risk assessment conclusion of the target patient based on the logical operation results of the rule engine module.

9. An electronic device, characterized in that, include: The device includes 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 communicates with the memory via the bus, and the processor executes the machine-readable instructions to perform the steps of the method as described in claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the method as described in claim 8.

Images