I-Corps: Translation potential of a clinical decision support system for evidence-based reasoning for disease management and prevention

This I-Corps project is based on the development of a counterfactual artificial intelligence (AI) technology that turns health data into understandable guidance about actions that may improve health outcomes. Patients and care teams increasingly face large amounts of continuous health data from wearable sensors, smartphones, and medical devices. However, most digital health tools only display past trends or warn of risk without clearly explaining what practical changes could help. The scale of this problem is growing as chronic conditions such as diabetes, cardiovascular disease, and related cardiometabolic disorders place major burdens on patients, families, clinicians, and healthcare systems. This technology analyzes patterns in health data to generate personalized “what-if” guidance about modifiable behaviors, treatments, or care strategies that may lead to better health outcomes. This technology may be used by healthcare systems, specialty clinics, remote patient monitoring platforms, digital health companies, and medical technology firms seeking tools that improve care efficiency and engagement. This may support earlier intervention, better self-management, reduced clinician burden, and broader access to specialty-informed guidance for chronic disease prevention and management. This I-Corps project utilizes experiential learning coupled with first-hand investigation of the industry ecosystem to assess the translation potential of counterfactual artificial intelligence (AI) for actionable clinical and health decision support. Existing digital health tools primarily visualize historical data or produce risk scores. This technology integrates multimodal longitudinal data streams, including wearable sensing, continuous physiological monitoring, connected medical-device data, and clinical records, to identify adverse events, forecast future risk, and generate patient-specific counterfactual explanations. These counterfactuals describe minimal, plausible changes in modifiable factors, such as behavior, treatment, or care strategy, that are predicted to improve a health trajectory. Previous research demonstrated methods for robust machine learning using wearable data, human-in-the-loop learning of health events, temporal modeling, synthetic time-series generation, and counterfactual frameworks for predicting and preventing adverse health events. This approach combines time-series forecasting, explainable AI, counterfactual reasoning, and human-centered decision support to connect prediction with action. This technology may transform complex longitudinal health data into transparent, intervention-oriented recommendations to improve health. 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.