MitPlan: A Planning Approach to Mitigating Concurrently Applied Clinical Practice Guidelines

Abstract

As the overall population ages, patient complexity and the scope of their care is increasing. Over 60% of the population over 65 years of age suffers from multi-morbidity, which is associated with over two times as many patient-physician encounters. Yet clinical practice guidelines (CPGs) are developed to treat a single disease. To reconcile these two competing issues, we developed a framework for identifying and addressing adverse interactions in multi-morbid patients managed according to multiple CPGs. The framework relies on first-order logic (FOL) to represent CPGs and secondary medical knowledge and FOL theorem proving to establish valid patient management scenarios. In this work, we leverage the framework’s representation capabilities to simplify its mitigation process and cast it as a planning problem represented using the Planning Domain Definition Language (PDDL). We demonstrate the framework’s ability to identify and mitigate adverse interactions using planning actions, add support for durative clinical actions, and show the improved interpretability of management plan recommendations in the context of both proof-of-concept and clinical examples.