NYU Langone Health developed an AI model to predict patients needing skilled nursing care post-hospitalization. ▪️ Analyzing 4,000 patients’ electronic health records, the AI identifies risk factors from admission notes. ▪️ The AI generates concise ‘AI Risk Snapshots’ summarizing discharge-relevant details. ▪️ Testing nine AI models, the model using short, AI-generated summaries proved most accurate. ▪️ The AI predictions aligned with human expert assessments, demonstrating 88% accuracy. ▪️ Early identification enables proactive care coordination, reduces discharge delays, and ensures appropriate post-hospital care. ▪️ Future steps involve real-world clinical testing for effectiveness, fairness, and safety monitoring.
AI tool predicts patients who will need care after hospital discharge
