Application of recommender system methods for therapy decision support

Abstract

In this paper two approaches for facilitating therapy decision support are proposed and compared. Both approaches, the Collaborative Recommender and hybrid Demographic-based Recommender, are based on recommender system methods which origin from the field of product recommendation in e-commerce applications. An exemplary dataset comprising health record excerpts of patients suffering from the skin disease psoriasis is used for evaluating both approaches. The approaches estimate the outcome of a subset of systemic therapies to support the medical practitioner in making therapy decisions for a specific patient and time, i.e. consultation under consideration. Both systems proved to work and are capable of assisting medical practitioners prospectively with making appropriate therapy decisions.