Adaptive mobile notifications generation based on physical activities

Abstract

Physical activity is an important factor that is considered for the prevention of diseases like diabetes or hypertension and for rehabilitation. Besides the advancement of technology and availability of smart-phones creates the opportunity to utilize the power of smartphone's sen- sors, for example, accelerometers, to support cost-e ective behavioral intervention to promote physical activities. In this thesis, we attempt to identify basic physical activities of a user from smartphone's 3D accelerometer data and then suggest the user through mobile phone noti cations the recommended level of physical activities he/she should undergo. In our work, we analyze an existing dataset containing accelerometer data with labeled physical activities, namely standing, walking, stair-up and stair-down and a few others, and learn the patterns identifying various activities. Once the patterns are learned, we identify series of activities that a certain user performs from its mobile phone accelerometer data determining what portion of time the user spends in what activities. Based on this information, we develop suggestions of performing activities for that user by analyzing his/her current and required amount of physical activities within a time window using some prede ned standard (amount of activities he/she must undergoes to be t and healthy). These suggestions are propagated to the user suggesting him/her to make further engagement in physical activities through mobile noti cation system.