Intelligent detection of pulse position in Traditional Chinese Medicine based on Bionic Pulse Diagnose Robotics

Abstract

Pulse position reflects the important physiological state of human body according to diagnosis theories of Traditional Chinese Medicine (TCM). Researchers in this field hope to develop an objective approach to detect pulse position without human intervention. In this study, we use our lab-developed bionic pulse diagnose robotics to automatically collect 10-egraedients pulsation waveforms for a total of 200 subjects. Meanwhile, the pulse position labelling of 0-100 is accomplished by a TCM specialist. We extract key features that are highly related to the pulse position according to TCM theories, and find that Summit Pulse Pressure, S1/S2, and Body Mass Index are significantly correlated with pulse position with R-values of -0.27, -0.29, -0.18, respectively. Then, we derive a pulse position prediction model based on the deep learning framework to automatically predict pulse position, with final prediction error of ± 10 at 93%, and error of ±5 at 83%. In summary, this study investigates pulse diagnosis robotics deriving key features and intelligent prediction model for pulse position, which laid a foundation for further research such as TCM big data and omics-wise investigations.