A Recommender System for Healthcare Based on Human-Centric Modeling

Personalized recommendation techniques are critical for a healthcare system to match patients with medical services. This work presented a human-centric user modeling approach and a recommendation system, which can work well at healthcare system with sparse operation data specially at the system's early stage. With human-centric modeling paradigm, both the user model and the item model are developed, and a matchmaking relationship are constructed between user intention and medical items. A user-to-user similarity matrix is developed to describe a user with other users. An item vector is developed to describe an item with its usage context. An experiment has been conducted on a healthcare prototype. The testing is encouraging and several proposed approaches and performance advantages are basically verified.