Exploring Factors Influencing Physician Selection Decisions

Selecting a physician from the physician review sites (PRS) is a challenging and complex task for patients. PRSs overload users with too many metrics, few ratings per doctor that are positively biased, and conflicts between numeric and text ratings. This paper addresses the issue of multiple ratings in the PRS application with analytical techniques of dimension reduction, such as PCA, and text mining, including Sentiment Analysis, to address conflicts between narrative and star rating. We collected and integrated patient review data from two leading PRS websites: Vitals and Healthgrades. Our findings indicate that PCA reduces the complexity of multiple rating metrics by reducing from seven metrics to one, while sentiment analysis alerts the patients to inconsistencies between numeric and narrative ratings, allowing patients to make accurate decisions. Additionally, we plan to develop an integrated dashboard website for patients to enable them to understand and use reviews from multiple websites efficiently