I-Corps: An AI-based Physician Advisory System for Disease Management

University Of Texas At Dallas, Richardson TX

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is its potential to help the 5.7 million adult Americans who are living with heart failure every year and potentially saving a significant fraction of the $30.7 billion dollars spent each year on heart failure care. It also provides a quantitative measure of the quality of heart failure care provided in hospitals and clinics. Optimal management of heart failure requires adherence to evidence based clinical guidelines. The Heart Failure Guidelines were created by a multi-disciplinary committee of experts and is based on thorough review of best available clinical evidence on management of heart failure. This guideline represents a consensus among experts on the appropriate treatment and management of heart failure. However, low adherence to the clinical guideline by physicians and cardiologists is one of the major challenges in the management of heart failure. This proposal seeks to overcome this challenge. The addressable market for this technology is about $106 million yearly in the United States alone with an even larger global market and the potential to develop additional extensions to manage similar diseases. This I-Corps project automates the entire set of rules in the current guideline for heart failure management. The system is based on answer set programming, a form of declarative programming suited for simulating human-style reasoning. Given a patient's information, the system generates a set of guideline-compliant recommendations. We conducted a pilot study of the system on 20 real and 10 simulated patients with heart failure. The results show that the system recommends treatments that are compliant with the guidelines for heart failure management. Out of 179 recommendations made by the system, expert cardiologists agreed upon 168 of them. The 11 recommendations that our system made not in agreement with cardiologists are due to misunderstanding of the doctor's notes or guideline which will be fixed in the future version of the software. Our research has established the foundation for automating the generation of guideline-compliant recommendations. Our plan is to develop the system into a point-of-care tool for clinic visits. The system would give real-time feedback of guideline-appropriate care recommendations, and be a quality assessment tool to measure the quality metrics of heart failure care of institutions. The same method applies to the management of other diseases such as Asthma and Cancer. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.