A predictive tool for nonattendance at a specialty clinic: An application of multivariate probabilistic big data analytics

The field of data analytics has recently garnered significant attention as a means of improving service, outcome and cost in healthcare. Health informatics tools to date are largely rules-based, static in knowledge-gathering capability, and are not well incorporated within standard clinical processes. We introduce a multivariate analysis tool, built at the user-client interface, directly from data extracted from thousands of EMRs, for automated decision making. The predictive algorithms created to date have been implemented in a variety of clinical scenarios, including ER triage of chest pain, selection of optimal diagnostic testing, and management of chronic illnesses. This technology has now been applied to prediction of patient compliance with scheduled appointments in a medical specialty clinic. Approximately 16% of scheduled VA patients do not show for their scheduled appointments for Gastroenterology clinic. This results in misallocation of provider resources and inefficiencies in health care delivery. The improvement in quality of care, patient outcomes and cost is potentially immense.