Causal Discovery of Health Features from Wearable Device and Traditional Chinese Medicine Diagnosis Data

This paper explores the cause-and-effect relationships among a set of health indices using causal discovery. The data we used to analyze was obtained from wearable devices, Traditional Chinese Medicine (TCM) diagnosis, and self-assessment of subjects in an experiment. Firstly, three machine learning algorithms were employed to address the issue of excessive missing values in the integrated dataset, and the coherence of this improved data was validated by statistical test. The NOTEARS algorithm was then employed to assess the causal relationships within the overall population as well as within subgroups based on gender, physical activity levels, and sleep duration. The results demonstrated that the NOTEARS algorithm yielded interesting and plausible outcomes, suggesting the presence of causal connections between variables of wearable devices and TCM diagnosis, as well as daily lifestyle habits.