Health Assessment for Office Workers by Tracking the Wrist Motion Centric Activity

Continuous monitoring and improving the health condition of office worker plays an important role in the holistic development of the organization. Very often many office workers overlook the proper diet and other health problems due to sedentary lifestyle and hectic office schedule. This paper presents a supervised learning-based solution for activity monitoring of office workers using wearable devices. The activity classification models were developed from a laboratory study consisting of five different activities such as drinking water, eating food, smoking, rubbing eyes and talking on the phone. The machine learning model is trained on the data collected from the wrist motion of subjects. The feature matrix is selected from the measured raw data of accelerometer, gyroscope, magnetometer in all three x, y and z-axis. For classification, we studied the performance of SVM, kNN and decision tree. The 5-fold cross-validation results in SVM having the best accuracy of 98.7% in the present scenario. Thus, our qualitative finding indicates the potential use of the smartphone-based wearable solution for the assessment of bad habits and improving the health condition during office work.