A location-sensitive over-the-counter medicines recommender based on tensor decomposition

Abstract

The last few decades have witnessed a steady increase in medicine prescriptions for the treatment of biometric markers rather than obvious physiological symptoms; especially, the over-the-counter (OTC) medicine experiences rated by patients have huge potential to assist people to make more appropriate decisions. The most existing researches focus on the rating prediction and recommendations in E-commerce field rather than healthcare or medical treatments. In addition, the spatial and temporal factors were not considered in their recommendation mechanisms. Toward this end, this paper propose an efficient OTC medicines recommendation strategy based on tensor decomposition. Considering the impact of regional differentiation, a third-order tensor including medicine, location, and rating is constructed. To inference the usage of a new OTC medicine in a certain location, high-order singular value decomposition is applied to the above tensor for obtaining the intelligent recommendation. In order to evaluate the effectiveness of the proposed approach, we compared the conventional collaborative filtering approach and tensor-based approach in terms of precision and recall. The experimental results demonstrate that our proposed approach is significant better than collaborative filtering approach.