AI-Driven Heart Failure Decision Support in Skilled Nursing Facilities

Heart failure management in skilled nursing facilities (SNFs) is complicated by limited access to specialists, incomplete clinical documentation, and patients with complex comorbidities. Artificial intelligence clinical decision support systems have been developed mostly for acute hospital settings but not for SNF settings. We present ForeSight-HF, an artificial intelligence clinical decision support tool designed specifically for SNF heart failure management. Built on a large language model with HIPAA-compliant infrastructure, ForeSight-HF synthesizes unstructured clinical documentation to generate heart failure likelihood scores, exacerbation risk scores, and individualized management recommendations. We illustrate its utility through 2 contrasting cases: a Warm/Wet (congested, well-perfused) patient for whom the system recommended intensified diuresis, and a Cold/Dry (hypovolemic, hypoperfused) patient for whom it recommended diuretic de-escalation to prevent harm. ForeSight-HF may help close a persistent care gap in postacute heart failure management.