Prescription Based Recommender System for Diabetic Patients Using Efficient Map Reduce

Abstract

Healthcare sector has been deprived of leveraging knowledge gained through data insights, due to manual processes and legacy record-keeping methods. Outdated methods for maintaining healthcare records have not been proven sufficient for treating chronic diseases like diabetes. Data analysis methods such as Recommendation System (RS) can serve as a boon for treating diabetes. RS leverages predictive analysis and provides clinicians with information needed to determine the treatments to patients. Prescription-based Health Recommender System (HRS) is proposed in this paper which aids in recommending treatments by learning from the treatments prescribed to other patients diagnosed with diabetes. An Advanced Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering is also proposed to cluster the data for deriving recommendations by using winnowing algorithm as a similarity measure. A parallel processing of data is applied using map-reduce to increase the efficiency & scalability of clustering process for effective treatment of diabetes. This paper provides a good picture of how the Map Reduce can benefit in increasing the efficiency and scalability of the HRS using clustering.