CN-122025081-A - Data processing method, device, computer equipment and medium based on artificial intelligence

CN122025081ACN 122025081 ACN122025081 ACN 122025081ACN-122025081-A

Abstract

The application belongs to the technical field of artificial intelligence, and relates to a data processing method, a device, computer equipment and a medium based on artificial intelligence, wherein the method comprises the steps of carrying out parallel reasoning on clinical input data based on a medical analysis model and a basic comparison model, and carrying out parallel reasoning on an obtained reasoning path and a basic reasoning path; the method comprises the steps of performing difference degree analysis and confusion degree analysis based on word dimensions to obtain difference degree data and confusion degree data, screening abnormal reasoning paths from reasoning paths and reference reasoning paths based on the confusion degree data if the difference degree data is larger than a preset threshold value, cutting abnormal contents of the abnormal reasoning paths to obtain stable reasoning segments, performing content generation processing on the stable reasoning segments and the designated reasoning paths based on a content generation model to obtain target reasoning results, and outputting the target reasoning results if the target reasoning results pass knowledge verification. The application can be applied to medical analysis scenes in the field of digital medical treatment, and improves the stability and reliability of medical analysis reasoning.

Inventors

WANG JIANZONG
QU XIAOYANG
ZHANG NAN

Assignees

平安科技（深圳）有限公司

Dates

Publication Date: 20260512
Application Date: 20260109

Claims (10)

1. A data processing method based on artificial intelligence, comprising the steps of: receiving clinical input data, and carrying out parallel reasoning on the clinical input data based on a preset medical analysis model and a basic comparison model to obtain a corresponding reasoning path and a reference reasoning path; performing word dimension-based difference degree analysis and confusion degree analysis on the reasoning path and the reference reasoning path to obtain corresponding difference degree data and confusion degree data; if the difference degree data is larger than a preset threshold value, screening out corresponding abnormal reasoning paths from the reasoning paths and the reference reasoning paths based on the confusion degree data; The abnormal content is cut off to the abnormal reasoning path to obtain a corresponding stable reasoning section; Performing content generation processing on the stable reasoning section and a designated reasoning path based on a preset content generation model to obtain a corresponding target reasoning result, wherein the designated reasoning path is the other path except the abnormal reasoning path in the reasoning path and the reference reasoning path; Carrying out knowledge verification on the target reasoning result based on a preset medical knowledge base; And if the target reasoning result passes the knowledge verification, outputting the target reasoning result.
2. The artificial intelligence based data processing method according to claim 1, wherein the step of screening out the corresponding abnormal inference path from the inference path and the reference inference path based on the confusion degree data specifically comprises: Acquiring specified confusion degree data of a first reasoning path, wherein the first reasoning path is any path in the reasoning path and the reference reasoning path; Analyzing the specified confusion degree data based on a preset confusion degree analyzer to judge whether the first reasoning path has an irregular reasoning mode or not; if the first reasoning path has an irregular reasoning mode, the first reasoning path is used as the abnormal reasoning path; And if the first reasoning path does not have the irregular reasoning mode, taking the first reasoning path as a normal reasoning path.
3. The artificial intelligence based data processing method according to claim 1, wherein the step of performing the truncation processing of the abnormal content on the abnormal reasoning path to obtain a corresponding stable reasoning segment specifically includes: Calling a preset logic consistency detector; Loading a preset medical knowledge base; scanning the abnormal reasoning path based on the logical consistency detector to identify abnormal content that does not meet relevant specifications of the medical knowledge base; cutting off the abnormal content in the abnormal reasoning path to obtain a processed second reasoning path; And taking the second reasoning path as the stable reasoning section.
4. The artificial intelligence-based data processing method according to claim 1, wherein the step of performing content generation processing on the stable inference segment and the specified inference path based on a preset content generation model to obtain a corresponding target inference result specifically comprises: extracting a target stable reasoning section from the appointed reasoning path based on a preset extraction strategy; integrating the stable reasoning section with the target stable reasoning section to obtain corresponding reasoning section data; Calling a preset content generation model; performing reasoning processing on the reasoning section data based on the content generation model to generate corresponding initial content; evaluating the initial content based on a preset evaluation strategy; And if the initial content passes the evaluation, taking the initial content as the target reasoning result.
5. The artificial intelligence based data processing method according to claim 4, wherein the step of performing the evaluation processing on the initial content based on the preset evaluation policy specifically comprises: Performing text similarity evaluation on the initial content based on the reasoning segment data; if the initial content passes the text similarity evaluation, a preset logic rule base is called; Performing logical relationship rationality evaluation on the initial content based on the logical rule base; If the initial content passes the logical relationship rationality evaluation, judging that the initial content passes the evaluation; and if the initial content does not pass the logical relationship rationality evaluation, judging that the initial content does not pass the evaluation.
6. The artificial intelligence-based data processing method according to claim 1, wherein the step of performing knowledge verification on the target inference result based on a preset medical knowledge base specifically comprises: analyzing the target reasoning result to obtain a corresponding analysis result; Calling a preset medical knowledge base; Comparing and verifying the analysis result with related knowledge in the medical knowledge base; if the analysis result passes the comparison verification, judging that the target reasoning result passes the knowledge verification; and if the analysis result fails the comparison verification, judging that the target reasoning result fails the knowledge verification.
7. The artificial intelligence based data processing method according to claim 1, further comprising, after the step of outputting the target inference result: collecting all relevant data corresponding to the reasoning generation process of the target reasoning result; report generation processing is carried out on all the related data to obtain a corresponding reasoning report; Acquiring a preset report output mode; And carrying out output processing on the reasoning report based on the report output mode.
8. An artificial intelligence based data processing apparatus comprising: The first processing module is used for receiving clinical input data, and carrying out parallel reasoning on the clinical input data based on a preset medical analysis model and a basic comparison model to obtain a corresponding reasoning path and a reference reasoning path; The analysis module is used for carrying out the difference degree analysis and the confusion degree analysis based on the word element dimension on the reasoning path and the reference reasoning path to obtain corresponding difference degree data and confusion degree data; the screening module is used for screening out corresponding abnormal reasoning paths from the reasoning paths and the reference reasoning paths based on the confusion degree data if the difference degree data is larger than a preset threshold value; The second processing module is used for carrying out cutting-off processing on the abnormal content of the abnormal reasoning path to obtain a corresponding stable reasoning section; The first generation module is used for carrying out content generation processing on the stable reasoning section and a designated reasoning path based on a preset content generation model to obtain a corresponding target reasoning result, wherein the designated reasoning path is the other path except the abnormal reasoning path in the reasoning path and the reference reasoning path; the verification module is used for carrying out knowledge verification on the target reasoning result based on a preset medical knowledge base; And the first output module is used for outputting the target reasoning result if the target reasoning result passes the knowledge verification.
9. A computer device comprising a memory having stored therein computer readable instructions which when executed implement the steps of the artificial intelligence based data processing method of any of claims 1 to 7.
10. A computer readable storage medium, characterized in that it has stored thereon computer readable instructions which, when executed by a processor, implement the steps of the artificial intelligence based data processing method according to any of claims 1 to 7.

Description

Data processing method, device, computer equipment and medium based on artificial intelligence Technical Field The application relates to the technical field of artificial intelligence, and can be applied to the field of digital medical treatment, in particular to a data processing method, a data processing device, computer equipment and a storage medium based on artificial intelligence. Background In the process of continuous advancement of medical construction, digital medical technology is vigorously developed, and a decision support system based on a large language model plays an increasingly important role in the medical field by virtue of the strong language understanding and generating capability of the decision support system, provides rich information and reference basis for medical decision, and is beneficial to improving the efficiency and quality of medical services. However, existing large language models encounter serious challenges in practical clinical applications, with the problem of inadequate analytical reasoning stability being particularly pronounced. The method is characterized in that the same large language model aims at similar cases, but an analysis reasoning process with inconsistent logic can be generated. This instability severely affects the clinician's confidence in the model output, reducing the reliability of the model in clinical applications, and thus limiting its effective play in medical decisions. For example, in the remote medical analysis scene of digital medical treatment, similar symptoms may exist in patients in different regions, when the existing large language model is used for auxiliary analysis, obvious logic differences exist in reasoning results such as etiology risk analysis and treatment proposal and the like given by the model for two cases with similar symptoms, so that doctors are difficult to accurately judge according to the model results, and the accuracy and reliability of remote medical analysis are affected. Therefore, the method has important practical significance for improving the reasoning stability of the large language model in medical application. Disclosure of Invention The embodiment of the application aims to provide a data processing method, a device, computer equipment and a storage medium based on artificial intelligence, so as to solve the technical problem that the analysis reasoning stability of the existing model is low in actual clinical application. In a first aspect, there is provided an artificial intelligence based data processing method, comprising: receiving clinical input data, and carrying out parallel reasoning on the clinical input data based on a preset medical analysis model and a basic comparison model to obtain a corresponding reasoning path and a reference reasoning path; performing word dimension-based difference degree analysis and confusion degree analysis on the reasoning path and the reference reasoning path to obtain corresponding difference degree data and confusion degree data; if the difference degree data is larger than a preset threshold value, screening out corresponding abnormal reasoning paths from the reasoning paths and the reference reasoning paths based on the confusion degree data; The abnormal content is cut off to the abnormal reasoning path to obtain a corresponding stable reasoning section; Performing content generation processing on the stable reasoning section and a designated reasoning path based on a preset content generation model to obtain a corresponding target reasoning result, wherein the designated reasoning path is the other path except the abnormal reasoning path in the reasoning path and the reference reasoning path; Carrying out knowledge verification on the target reasoning result based on a preset medical knowledge base; And if the target reasoning result passes the knowledge verification, outputting the target reasoning result. In a second aspect, there is provided an artificial intelligence based data processing apparatus comprising: The first processing module is used for receiving clinical input data, and carrying out parallel reasoning on the clinical input data based on a preset medical analysis model and a basic comparison model to obtain a corresponding reasoning path and a reference reasoning path; The analysis module is used for carrying out the difference degree analysis and the confusion degree analysis based on the word element dimension on the reasoning path and the reference reasoning path to obtain corresponding difference degree data and confusion degree data; the screening module is used for screening out corresponding abnormal reasoning paths from the reasoning paths and the reference reasoning paths based on the confusion degree data if the difference degree data is larger than a preset threshold value; The second processing module is used for carrying out cutting-off processing on the abnormal content of the abnormal reasoning path to obtain a correspondi