Determining Important Features in Multidimensional Health Data for Individualized Precision Healthcare

In recent years, there has been a growing need for individuals' health management by using sensors and wearable devices to record daily activity and monitor health indicators. A large amount of health data needs to be analyzed to investigate the essential impact factors related to individuals' health and help individuals manage their health. In this paper, we investigate the important features influencing personal health from health data obtained from wearable devices and health data based on Traditional Chinese Medicine (TCM). We focus on investigating and selecting more influential health features and then performing machine learning algorithms for modeling. The results show that daily activity consumption is of a greater influence on wearable device data, and the pulse position that represents the kidney is identified as having the most significant impact on TCM health status among all pulse positions. Moreover, we selected the most influential features to perform the regression model and compared them with all the features. The results show that after feature selection, the Mean Squared Error (MSE) is smaller, and the R-square Score (R2) is greater than before.