Rare Disease Detection and Physician Targeting: A Factor Graph Machine Learning Approach for Niche Market Targeting

Abstract

A rare disease is any disease that affects a small percentage of the population. The extremely low incidence rate of rare diseases makes it particularly difficult to recognize and diagnose them. A major challenge in the rare disease market is how to target physicians who are potentially involved with patients with a specific rare disease. The existing detection and targeting methods, such as segmentation and profiling, have been developed under an assumption of a large mass market and are thus not suitable for the rare disease market where the classification classes are extremely imbalanced. This paper proposes a factor graphical model approach to predict rare disease physician targets by jointly modeling physician and patient features from different data spaces and explicitly incorporating physician-patient relationship. Through an empirical application in detecting physicians treating hereditary angioedema using big medical claims and prescription data, the proposed approach demonstrates better performances than various benchmark models according to different performance metrics. The graph representation also allows for visual interpretation of the relationship between physicians and patients. This paper contributes to the literature on exploring the benefits of utilizing relational dependencies among entities in healthcare industry.

Keywords: factor graph, machine learning, healthcare, imbalanced class, rare disease, physician targeting, disease detection