Perspective Chapter: Highly Structured Data Collection and Predictive Diagnostic Assist Algorithms in Orofacial Pain and Oral Medicine

Abstract

We have constructed a highly structured note-taking system that is readily minable because it consists of objective data rather than free text and is free of identifiable, protected health information. Our objective observations automatically enter a backend database collected from patients who have attended our clinic for orofacial pain and oral medicine. We next added two Naive Bayesian algorithms into our encounter process to generate real-time diagnostic suggestions for each patient. The first is a standard Naive Bayesian Inference algorithm that provides the clinician with a dynamic, real-time list of possible diagnoses that the collected data support. The second is a Naive Bayesian Exploration algorithm that provides the clinician with a second list of the critical data that should be collected next to confirm or refute the suggested diagnoses. We are now comparing the algorithm-assisted note-taking system to notes created without the Bayesian algorithms. Our planned outcomes are reduced click burden and concordance between the clinician’s diagnosis and those predicted by the algorithms. This project aims to examine if an algorithm-based clinical decision-support system improves the efficiency and accuracy of the diagnostic process.