Personalized Body Constitution Inquiry Based on Machine Learning.

Area of Science:

• Integrative Medicine
• Artificial Intelligence in Healthcare
• Traditional Chinese Medicine (TCM)

Background:

• Body constitution (BC) in TCM reflects a person's health status, traditionally assessed by doctors through physical indicators and inquiry.
• Existing methods often lack subjective patient feelings, impacting diagnostic accuracy.
• The Constitution in Chinese Medicine Questionnaire (CCMQ) collects subjective data but is inefficient.

Purpose of the Study:

• To develop a personalized body constitution inquiry method using machine learning.
• To enhance the accuracy and efficiency of BC type identification.
• To simulate a doctor's diagnostic inquiry process for TCM.

Main Methods:

• A machine learning model incorporating a random generator, feature extractor, and classifier was developed.
• The feature extractor prioritizes CCMQ questions based on tongue coating image analysis.
• A classifier screens questions to generate a personalized questionnaire for each patient.

Main Results:

• The personalized method significantly reduced the number of questions by 68.3% and answering time by 80.3% compared to CCMQ.
• Improved accuracy in recognizing body constitution types was achieved.
• The system successfully screened personalized questions for patients.

Conclusions:

• The proposed method effectively simulates a doctor's inquiry, providing personalized questions for BC assessment.
• This tool aids in collecting subjective patient data for more accurate BC diagnosis.
• It serves as a valuable auxiliary diagnostic tool for TCM practitioners.