Training method of syndrome diagnosis model and syndrome prediction method and device

By conducting multi-level training and pseudo-label correction on a TCM corpus and a database of medical records from renowned veteran TCM doctors, the accuracy and interpretability of the syndrome diagnosis model were improved, solving the problem of high data annotation complexity in TCM syndrome differentiation and diagnosis.

CN122117264APending Publication Date: 2026-05-29PEKING UNIV +1
0 Cites 0 Cited by

Patent Information

Application Number
CN202610112957.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-27
Publication Date
2026-05-29

AI Technical Summary

Technical Problem

Existing technologies in TCM syndrome differentiation and diagnosis suffer from limited data volume, inconsistent terminology, and complex syndrome element labeling, which limits the accuracy and interpretability of intelligent syndrome diagnosis models.

Method used

Unsupervised training was conducted using a basic TCM corpus, and supervised training was conducted using a database of general and specific medical records of renowned veteran TCM doctors. Pseudo-labels for syndrome elements were generated and corrective measures were taken to optimize the syndrome diagnosis model.

Benefits of technology

It improves the accuracy and interpretability of syndrome differentiation and diagnosis, reduces the complexity of data annotation, and enables in-depth understanding and accurate simulation of the syndrome differentiation and treatment logic of renowned traditional Chinese medicine practitioners.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a syndrome diagnosis model training method and a syndrome prediction method and device. The syndrome diagnosis model training method comprises the steps of performing unsupervised training on an initial model through a basic traditional Chinese medicine corpus to obtain a first diagnosis model; supervised training is conducted on the first diagnosis model through a general famous old traditional Chinese medicine case database, a second diagnosis model is obtained, and the general famous old traditional Chinese medicine case database comprises syndrome element information corresponding to the symptoms of the patient; determining a syndrome element pseudo tag corresponding to the specific famous old traditional Chinese medicine case database, and performing correction processing on the syndrome element pseudo tag to obtain a target pseudo tag; and performing optimization training on the second diagnosis model through the specific famous old traditional Chinese medicine case database and the corresponding target pseudo tag to obtain a syndrome diagnosis model. According to the method and the device, the technical problems of limited accuracy and insufficient interpretability of an intelligent syndrome diagnosis model caused by deficiency of syndrome element labeling and high labeling difficulty in famous and old traditional Chinese medicine case data are solved.
Need to check novelty before this filing date? Find Prior Art