ABSTRACT
This study was aimed to explore the feasibility of identification of traditional Chinese medicine (TCM) constitution based on photoplethysmography (PPG) frequency domain analysis,and to establish an automatic identification system (AIS) of TCM constitution with simple operation and good generalization.The Classification and Determination of Constitution in TCM was used to evaluate the subjects' constitutional state.Pulse wave sensor was used to collect the pulse wave signal.Spectral characteristics were analyzed by power spectrum analysis.Eight classification algorithms of data mining were used to establish the model of qi-insufficiency constitution.The results showed that the pulse wave harmonic peak was that qi-insufficiency constitution was higher than health group;the model was optimal when random forest algorithm was used;and the accuracy of 10-CV test was 80.4％.Sensitivity reached 73.2％.Specificity and receiver operator characteristic were all above 0.8.The model has certain reliability.It was concluded that it was feasible to identify qiinsufficiency constitution by means of frequency domain analysis of PPG.It provides preliminary work for the construction of AIS of TCM constitution as well as a simple and quick assistant tool for the constitution identification of subhealth group in family health care.