International Business Machines Corporation
Personalized activity adviser model

Abstract:

A processor(s) receives a profile of a user and ranks physical activities in a relational data structure based on the user profile to create a customized relational data structure of ranked physical activities for the user. The relational data structure maps physical activities to one or more associated attributes. An activity adviser model is produced specific to the user via machine learning by obtaining data related to physical activities performed by the user from one or more sensors, and receiving user wellness-related feedback related to the user-performed physical activities. The user wellness-related feedback includes a user ranking of one or more aspects of the user-performed physical activities. The activity adviser model specific to the user is built by using the customized relational data structure, the data and the user wellness-related feedback. The processor(s) uses the activity adviser model to provide an activity-related recommendation to the user.