Measuring the Efficacy of an Artificial Intelligence Predictive Algorithm that Reduces Misdiagnosis of Chronic Diseases

Background: Medical errors and health-care related adverse events occur in 8% to 12% of hospitalizations and misdiagnosis of chronic conditions in particular is one of the leading causes of death. A leading cause is that critical patient and diagnostic information flows neither consistently, nor timely, between specialties and institutions. Chronic disease patients often have multiple concurrent conditions and complications that require different specialists, while the needs of aging populations add complexity to their care: Diverse patients need multiple and different varied care specialties, each with its own clinical workflow.

Methods: A potential solution is based upon a new software technology to improve specialist workflow and communication. The authors are developing artificial intelligence (AI) predictive diagnosis algorithms to study data generated which include; patient history, measurements, doctor’s reports, and outcomes and compare to each specialist’s diagnosis patterns. The predictive algorithm recommends treatment alternatives, identifies patients who are at high risk for other conditions than the ones they were tested for, identifies high-relevance candidates for drug/treatment trials, and identifies areas for research on cause/effect, based on patient commonalities in history and outcomes.

Results: The authors are testing the hypothesis that key performance indicators to assess the efficacy of the AI algorithm include: the periods over which readmission rates and outcomes compared with patient diagnoses and treatments received; for cardiovascular diseases matched with other chronic diseases under initial focus.

Conclusion: The authors are pursuing the goal of improving patient outcomes through the use of their technology. It is the authors’ intention that through this talk, applicability to matching other conditions will be identified.