Classification of Patients’ Judgments of Their Physicians in Web-Based Written Reviews Using Natural Language Processing

Patients increasingly rely on web-based physician reviews to choose a physician and share their experiences. However, the unstructured text of these written reviews presents a challenge for researchers seeking to make inferences about patients’ judgments. This study aims to train, test, and validate an advanced natural language processing algorithm for classifying the presence and valence of 2 dimensions of patient judgments in web-based physician reviews: interpersonal manner and technical competence.

Basically: After scraping around 100,000 reviews patients posted after seeing their physicians, we wanted to figure out when/whether they were commenting on whether they thought the doctor (1) had good/bad bedside manner and/or (2) was technically competent–good at diagnosis, skilled at a procedure, etc. In this paper, we show that we successfully trained an algorithm to figure this out for us. The study was led by Farrah Madanay, a former Public Policy PhD student at Duke. To see more, go here: