A Modified Bayesian Model for Stress Testing: Potential Use in Artificial Intelligence Systems

Bayes’ theorem has been an integral part of traditional treadmill stress testing for many years. However, several issues have been raised regarding this approach: imperfect identification of important coronary obstruction, limited usefulness in certain patients, inconsistent results and interpretations, and confusion regarding the outcomes of various studies. This has led to disenchantment with basic stress testing in favor of more expensive methods, and difficulties for the clinician in managing patients. The proposal presented in this paper modifies the traditional Bayes approach so as to provide greater clinical utility for physicians performing treadmill stress tests. The model has potential for use in artificial intelligence (AI) systems.