Developing a Healthcare Recommender System Using an Enhanced Symptoms-Based Collaborative Filtering Technique

摘要

Recommendation system techniques are widely applied in different Web-based applications such as recommending movies, music, news, books, research articles, social tags, and products. Healthcare recommender systems (HRS) provide medical information based on patients' personal health record (PHR). The present research modifies the features of the traditional collaborative filtering technique to be used in a new symptoms-based collaborative filtering (SCF) approach. In the proposed method, symptoms and diseases are used instead of items and users respectively. The key advantage of this developed technique underlies its ability to recommend the possible emergence of diseases to health authorities. The SCF technique was applied based on social data that was collected from social media networks, Twitter. The collected data were classified into positive and negative groups. The former refers to patients, whereas the latter indicates that users are non-patients. The findings support the effectiveness of the proposed approach in comparison to real statistics obtained from the reports of the World Health Organization (WHO). The similarity between the proposed approach and the WHO data reached up to 98%. Moreover, the SCF approach can predict the disease one or two weeks in advance to health authorities.