System and method for generating medical analysis report

Abstract

The invention provides a method and system for generating a medical analysis report, and the method comprises the steps: receiving text data, extracting first sub-data from the text data based on a natural language processing technology through a disassembly module, analyzing the first sub-data through at least one medical database and a rule engine to generate a first intermediate result, executing the large language model to generate and output a first medical analysis report according to the first intermediate result, and disassembling the first medical analysis report or the text data into second sub-data by utilizing a disassembling module and feedback information responding to the first medical analysis report, and analyzing the second sub-data by using the medical database and the rule engine to generate a second intermediate result, and executing the large language model to generate and output a second medical analysis report according to the second intermediate result.

Description

Technical Field

[0001] The present invention relates to data analysis, and in particular to a system and method for generating a medical analysis report. Background Art

[0002] When a patient is admitted to the hospital, medical staff need to manually write a series of important files, including admission notes (AN), admission orders (AO) and progress notes (PN). These files are crucial for the treatment and management of patients, but medical staff often face multiple challenges when generating these files.

[0003] First, writing these records is time-consuming and labor-intensive. It is a tedious and time-intensive process for medical staff to record the patient's medical history, physical examination results, diagnosis and treatment plan in detail. Second, this paperwork often prevents medical staff from investing more time in direct patient care. Third, these records need to be revised many times. As the patient's condition changes, medical staff need to constantly update the medical records to reflect the latest diagnosis and treatment plan. This causes medical staff to frequently return to paperwork, increasing the workload. Summary of the invention

[0004] In view of this, the present invention proposes a system and method for generating a medical analysis report to solve the above-mentioned problem.

[0005] A method for generating a medical analysis report according to one embodiment of the present invention includes executing, using a computing device: receiving text data, extracting at least one first sub-data from the text data using a decomposition module based on natural language processing technology, analyzing at least one first sub-data using at least one medical database and at least one rule engine to generate at least one first intermediate result, executing a large language model to generate and output a first medical analysis report based on the at least one first intermediate result, decomposing the first medical analysis report or text data into at least one second sub-data using the decomposition module and in response to feedback information of the first medical analysis report, analyzing at least one second sub-data using at least one medical database and at least one rule engine to generate at least one second intermediate result, and executing a large language model to generate and output a second medical analysis report based on the at least one second intermediate result.

[0006] According to an embodiment of the present invention, a system for generating a medical analysis report includes an input device, a storage device, a computing device, and an output device. The input device is used to receive text data, a first medical analysis report, and feedback information. The storage device is used to store a disassembly module, at least one medical database, at least one rule engine, and a large language model. The computing device is communicatively connected to the input device and the storage device. The computing device uses the disassembly module to extract at least one first sub-data from the text data based on natural language processing technology, uses at least one medical database and at least one rule engine to analyze at least one first sub-data to generate at least one first intermediate result, executes the large language model to generate and output the first medical analysis report based on the at least one first intermediate result, uses the disassembly module and the feedback information in response to the first medical analysis report to disassemble the first medical analysis report or the text data into at least one second sub-data, uses at least one medical database and at least one rule engine to analyze at least one second sub-data to generate at least one second intermediate result, and executes the large language model to generate and output the second medical analysis report based on the at least one second intermediate result. The output device is communicatively connected to the computing device to output the first medical analysis report and the second medical analysis report.

[0007] In summary, the system and method for generating medical analysis reports proposed by the present invention have deep data analysis capabilities and are good at interpreting complex medical records, just like a general practitioner accurately summarizing and analyzing symptoms in sequence; secondly, the system and method proposed by the present invention have quality control guarantees, which can identify potential problems in medical reports, reduce human errors and negligence, and ensure the highest quality patient care information. In addition, the system and method proposed by the present invention can realize real-time assistance in medical care, automate the execution of daily medical paperwork, and allow the team to share consistent and real-time updated patient records.

[0008] The above description of the content of the present invention and the following description of the embodiments are used to demonstrate and explain the spirit and principle of the present invention, and to provide further explanation of the claims of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram of a system for generating a medical analysis report according to an embodiment of the present invention;

[0010] Figure 2 It is a data flow graph of multiple operations of a software system running on a computing device;

[0011] Figure 3 is a diagram of the internal architecture of the analysis module; and

[0012] Figure 4 is a flow chart of a method for generating a medical analysis report according to an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The detailed features and advantages of the present invention are described in detail in the following embodiments, and the contents are sufficient to enable any person skilled in the art to understand the technical content of the present invention and implement it accordingly, and according to the contents disclosed in this specification, claims and drawings, any person skilled in the art can easily understand the relevant purposes and advantages of the present invention. The following examples further illustrate the viewpoints of the present invention in detail, but do not limit the scope of the present invention in any viewpoint.

[0014] Figure 1 FIG. 1 is a schematic diagram of a system for generating a medical analysis report according to an embodiment of the present invention. Figure 1 As shown, the system 10 for generating a medical analysis report includes an input device 1 , a storage device 3 , a computing device 5 , and an output device 7 .

[0015] The input device 1 is used to receive text data, the first medical analysis report and feedback information. In one embodiment, the input device 1 is a hardware component such as a keyboard, a mouse or a touchpad. In another embodiment, the input device 1 is a software component such as an application programming interface (API) or a database, but the present invention is not limited to the above examples. The source of the text data can be direct input by the user, or text converted from voice or image.

[0016] The storage device 3 is used to store the disassembly module 50, at least one medical database, at least one rule engine and a large language model. In one embodiment, the storage device 3 is, for example, a flash memory, a hard disk (HDD), a solid state drive (SSD), a dynamic random access memory (DRAM), a static random access memory (SRAM) or other non-volatile memory, but the present invention is not limited to the above examples.

[0017] The computing device 5 is connected to the input device 1 and the storage device 3. The computing device 5 is used to run the software system proposed by the present invention, eXpertMind TMEngine. This software system includes the following multiple operations: using a disassembly module to extract at least one first sub-data from text data based on natural language processing technology, using at least one medical database and at least one rule engine to analyze at least one first sub-data to generate at least one first intermediate result, executing a large language model to generate and output a first medical analysis report based on at least one first intermediate result, using a disassembly module and responding to feedback information of the first medical analysis report to disassemble the first medical analysis report or text data into at least one second sub-data, using at least one medical database and at least one rule engine to analyze at least one second sub-data to generate at least one second intermediate result, and executing a large language model to generate and output a second medical analysis report based on at least one second intermediate result. In one embodiment, the computing device 5 may adopt at least one of the following examples: a personal computer, a network server, a central processor unit (CPU), a graphics processing unit (GPU), a neural network processing unit (NPU), a microcontroller (MCU), an application processor (AP), a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system-on-a-chip (SOC), a deep learning accelerator, or any electronic device with similar functions. The present invention does not limit the hardware type of the computing device 5.

[0018] The output device 7 is communicatively connected to the computing device 5 to output the first medical analysis report and the second medical analysis report. In one embodiment, the output device 7 is, for example, a screen, a projector, a speaker, or any hardware component or software component (such as an API interface or a database) that provides graphics and text, and the present invention is not limited thereto.

[0019] Figure 2 is a data flow diagram of multiple operations of the software system running on the computing device 5. Figure 2As shown, the software system 40 includes a disassembly module 50, analysis modules 61, 62 and 63, and a large language model 70. The present invention is characterized in that the medical analysis report can be re-generated. In other words, if the medical analysis report D4 generated in the first round is not satisfactory, it can be sent back to the software system 40, and the software system 40 generates a new medical analysis report based on the medical analysis report D4 and the feedback information D5 of the previous round.

[0020] The first round of operation of the software system 40 is as follows: the user generates text data D1 through the input device 1, and the decomposition module 50 decomposes the text data D1 into a plurality of sub-data D21, D22 and D23. The analysis modules 61, 62 and 63 use the built-in medical knowledge base and rule engine to analyze the sub-data D21, D22 and D23 to generate a plurality of intermediate results D31, D32 and D33. The large language model 70 (LLM) generates a medical analysis report D4 based on the intermediate results D31, D32 and D33. The second round of operation flow of the software system 40 is as follows: the medical analysis report D4 generated in the first round is re-input into the software system 40, and the user can selectively give feedback information D5 based on the medical analysis report D4. The disassembly module 50 then disassembles it into multiple new sub-data based on the feedback information D5 and / or the text data D1. The analysis modules 61, 62 and 63 then analyze these new sub-data to generate new intermediate results. Finally, the large language model 70 generates a new medical analysis report based on the new intermediate results.

[0021] The text data D1 is, for example, an admission record drafted by a professional, and the content includes, for example, but not limited to, the chief complaint, present illness, review of systems, physical examination, and clinical laboratory result. The medical analysis report D4 generated for the first time is, for example, the initial overall analysis report (OA). The feedback information D5 includes the user's score and prompt for the medical analysis report D4. The score is used to adjust the analysis process of the software system 40 and improve the quality of the new medical analysis report generated in the next round. The software system 40 can record the user's question type, question content, and feedback information D5. A high score indicates that the user is satisfied with the suggestions generated by the software system 40, so the weight of the current analysis strategy will be increased for the same type of question next time. A low score indicates that the result fails to meet the user's needs or expectations, so the weight of the current analysis strategy will be reduced for the same type of question next time, and different parameter settings will be tried to adjust the analysis strategies of the analysis modules 61, 62, and 63 to generate more appropriate results. The prompt word is used by the user to instruct the software system 40 to generate a new format of the medical analysis report D4 or to continue to use the existing format of the medical analysis report D4. 1 Afterwards, the user can use the prompt word to instruct the software system 40 to repeatedly generate a new OA 2 , OA 3 ,…,OA N The user can also use the prompt word to instruct the software system 40 to generate a medical analysis report D4 in other formats, such as an admission record AN, an admission doctor's order AO, or a medical history record PN. In addition, the process of generating a medical analysis report D4 in a new format can also be repeated. In other words, after generating an AN 1 / AO 1 / PN 1 Afterwards, if the user is not satisfied with the content, he can continue to generate a new AN in the same way as before. 2 / AO 2 / PN 2 .

[0022] The disassembly module 50 uses natural language processing (NLP) technology to understand and analyze the user's input. NLP technology can identify keywords, semantic structures and intentions.

[0023] Figure 3 This is the internal architecture diagram of the analysis module (taking 61 as an example), such as Figure 3As shown, the analysis module 61 has multiple built-in rule engines 81, 82...8N and multiple medical databases 91, 92...9N. The rule engines 81, 82...8N are based on the keywords, semantic structures and data results identified by the disassembly module 50, in conjunction with the large amount of medical data and expert knowledge recorded in the medical databases 91, 92...9N, and adjust the weight of the analysis strategy according to different feedback information D5, thereby running an expert system that can simulate the thinking logic of professional physicians, and performing corresponding analysis on multiple projects, including but not limited to: differential diagnosis, information to be further understood, the most likely diagnosis, host factors, inferred pathogenic mechanisms, other diagnostic examinations, imaging examinations, the need for emergency treatment, drug treatment, potential complications, preventive measures after management, etc.

[0024] In one embodiment, the large language model 70 and the medical database generate reference data as a part of the intermediate results D31 , D32 and / or D33 based on the Retrieval Augmented Generation (RAG) technology.

[0025] Tables 1, 2 and 3 below are used for illustration purposes only. Figure 2 and Figure 3 The specific forms of the text data D1, sub-data D21, D22 and D23, and intermediate results D31, D32 and D33 are shown in Table 3, where Table 3 is also used to present the process of the rule engine operation, but the present invention is not limited to this example.

[0026] Table 1, example of text data D1.

[0027]

[0028] Table 2, example of sub-data D21.

[0029] type project Numeric mark Basic data gender male Basic data Date of Birth 1967 / 10 / 10 Basic data age 49 Personal medical history diabetes Positive Personal medical history High cholesterol Positive Personal medical history hypertension Positive Personal medical history Hepatitis B Positive Carrier Laboratory tests GOT 190 U / L Laboratory tests GPT 149 U / L Laboratory tests albumin 39 g / L Laboratory tests globulin 40 g / L Laboratory tests γ-GT 309 U / L Laboratory tests Direct bilirubin 20.52 umol / L Laboratory tests Indirect bilirubin 20.52 umol / L Laboratory tests Total bilirubin 41.04 umol / L Laboratory tests Alkaline phosphatase 668 U / L Laboratory tests IgG anti-HAV 2.20(+) S / CO Laboratory tests HBsAg 7411(+) Index Laboratory tests Anti-HCV 0.1(-) Index

[0030] Table 3, Operation example of rule engine and data database.

[0031]

[0032] Figure 4The flowchart of the method for generating a medical analysis report according to an embodiment of the present invention includes steps S1 to S7. In step S1, the computing device 5 receives text data from the input device 1. In step S2, the computing device 5 uses the disassembly module 50 to extract at least one first sub-data from the text data based on natural language processing technology. In step S3, the computing device 5 uses at least one medical database and at least one rule engine to analyze at least one first sub-data to generate at least one first intermediate result. In step S4, the computing device 5 executes the large language model 70 to generate and output a first medical analysis report based on at least one first intermediate result. In step S5, the computing device 5 uses the disassembly module 50 and the feedback information in response to the first medical analysis report to disassemble the first medical analysis report or the text data into at least one second sub-data. In step S6, the disassembly module 50 uses at least one medical database and at least one rule engine to analyze at least one second sub-data to generate at least one second intermediate result. This step includes adjusting the weights used in the analysis according to the feedback information to generate the at least one second intermediate result, wherein the weights correspond to the acceptability in the feedback information, and the at least one second intermediate result is different from the at least one first intermediate result. In step S7, the computing device 5 executes the large language model 70 to generate and output a second medical analysis report based on the at least one second intermediate result. In the above process, it may also include: the computing device 5 generates reference data as part of the at least one first intermediate result based on the retrieval enhancement generation technology using the large language model 70 and at least one medical database.

[0033] The following describes an application scenario of the system and method for generating a medical analysis report proposed by the present invention. First, a medical professional provides text data to the system 10 for generating a medical analysis report proposed by an embodiment of the present invention. Then, the software system 40 running on the computing device 5 performs an initial analysis based on the text data, comprehensively evaluates the clinical condition of the patient, and generates a first original analysis (OA). 1 ). If OA 1 If the result is not satisfactory, you can enter feedback information and prompt words, and 1 The feedback information D5 is submitted to the software system 40, and then the software system 40 uses the text data and the previous OA 1 Generate new OA 2 This process can be repeated as many times as needed to produce OA 3 OA 4 ,…,OA X Then, the professional selects the most accurate one among these original analyses, such as OA i , is re-entered into the software system 40 as input data, and a prompt word and a text data question are attached, wherein the prompt word requires the provision of an admission record (AN1 ) and Admission Instructions (AO) 1 ). If the first AN 1 and AO 1 If the result is not satisfactory, you can enter feedback information, prompt words, original questions, and 1 , AO 1 The above information is submitted to the software system 40, and then the software system 40 generates a new AN 2 and OA 2 This process can be repeated as many times as needed to generate multiple groups of AN 3 and OA 3 , AN 4 and OA 4 ,…,AN x and OA x Ultimately, medical professionals select one of these admission records and admission instructions and revise its content as the final version.

[0034] The above application scenarios can be changed to generate medical records (PN), cancer treatment strategies, medical question responses, health examination reports, outpatient SOAP analysis (Subjective, Objective, Assessment, Plan), nursing report analysis, department examination reports, etc. By repeatedly inputting the previous output results and feedback information into the software system 40, a medical analysis report that better meets the needs of medical personnel can be obtained.

[0035] In summary, the system and method for generating medical analysis reports proposed by the present invention have deep data analysis capabilities and are good at interpreting complex medical records, just like a general practitioner accurately summarizing and analyzing symptoms in sequence; secondly, the system and method proposed by the present invention have quality control guarantees, which can identify potential problems in medical reports, reduce human errors and negligence, and ensure the highest quality patient care information. In addition, the system and method proposed by the present invention can realize real-time assistance in medical care, automate the execution of daily medical paperwork, and allow the team to share consistent and real-time updated patient records.

[0036]

Explanation of symbols

[0037] 1: Input device

[0038] 3: Storage device

[0039] 5: Computing device

[0040] 7: Output device

[0041] 10: System for generating medical analysis reports

[0042] 40: Software System

[0043] 50: Disassembly module

[0044] 61,62,63: ​​Analysis module

[0045] 70: Large Language Model

[0046] 81,82,8N: Rule Engine

[0047] 91,92,9N: Medical database

[0048] D1: text data

[0049] D21, D22, D23: Sub-data

[0050] D31, D32, D33: Intermediate results

[0051] D4: Medical analysis report

[0052] D5: Feedback Information

[0053] S1-S7: steps.

Claims

1. A method for generating a medical analysis report, characterized in that: The method comprises executing, by a computing device: Receive text data; Extracting at least one first sub-data from the text data using a disassembly module based on a natural language processing technology; Analyzing the at least one first sub-data using at least one medical database and at least one rule engine to generate at least one first intermediate result; executing a large language model to generate and output a first medical analysis report according to the at least one first intermediate result; Decomposing the first medical analysis report or the text data into at least one second sub-data by using the decomposition module and the feedback information in response to the first medical analysis report; Analyzing the at least one second sub-data using the at least one medical database and the at least one rule engine to generate at least one second intermediate result; as well as The large language model is executed to generate and output a second medical analysis report according to the at least one second intermediate result.

2. The method for generating a medical analysis report according to claim 1, wherein: Also includes: The large language model and the at least one medical database are used to generate reference data as a part of the at least one first intermediate result based on the search enhancement generation technology.

3. The method for generating a medical analysis report according to claim 1, wherein using the at least one medical database and the at least one rule engine to analyze the at least one second sub-data to generate the at least one second intermediate result comprises: The weight used in the analysis is adjusted according to the feedback information to generate the at least one second intermediate result. 4 . The method for generating a medical analysis report according to claim 3 , wherein the weight corresponds to an acceptability level of the feedback information. 5 . The method for generating a medical analysis report according to claim 1 , wherein the at least one second intermediate result is different from the at least one first intermediate result.

6. A system for generating a medical analysis report, characterized in that: include: An input device for receiving text data, a first medical analysis report and feedback information; A storage device for storing the disassembly module, at least one medical database, at least one rule engine and a large language model; A computing device, communicatively connected to the input device and the storage device, the computing device extracting at least one first sub-data from the text data using a decomposition module based on a natural language processing technique, analyzing the at least one first sub-data using the at least one medical database and the at least one rule engine to generate at least one first intermediate result, executing a large language model to generate and output a first medical analysis report based on the at least one first intermediate result, decomposing the first medical analysis report or the text data into at least one second sub-data using the decomposition module and feedback information in response to the first medical analysis report, analyzing the at least one second sub-data using the at least one medical database and the at least one rule engine to generate at least one second intermediate result, and executing the large language model to generate and output a second medical analysis report based on the at least second intermediate result; as well as The output device is communicatively connected to the computing device to output the first medical analysis report and the second medical analysis report.

7. The system for generating a medical analysis report according to claim 6, wherein the computing device is further used to generate reference data as a part of the at least one first intermediate result based on the retrieval enhancement generation technology using the large language model and the at least one medical database.

8. The system for generating a medical analysis report according to claim 6, wherein using the at least one medical database and the at least one rule engine to analyze the at least one second sub-data to generate the at least one second intermediate result comprises: The weight used in the analysis is adjusted according to the feedback information to generate the at least one second intermediate result. 9 . The system for generating a medical analysis report according to claim 8 , wherein the weight corresponds to an acceptability level of the feedback information. 10 . The system for generating a medical analysis report according to claim 6 , wherein the at least one second intermediate result is different from the at least one first intermediate result.

Images