How AI Is Predicting Hospital Readmissions Within 24 Hours

Episode 45 of Healthtech Talks with Fexingo. Lucas and Luna dive into a new wave of AI models that can predict which patients are likely to be readmitted to the hospital within 30 days of discharge — and do it in under 24 hours after admission. They examine a recent study from Mount Sinai Health System in New York, where a deep-learning algorithm trained on electronic health record data achieved an area under the curve of 0.86 — outperforming the traditional LACE index by about 12 percentage points. Lucas explains the three data streams the model uses: vital sign trajectories, medication changes, and nursing notes. Luna questions whether these tools actually change clinician behavior. The episode also explores the operational challenge of turning a prediction into an intervention — like scheduling a follow-up appointment before the patient leaves the bed. Relevant for hospital administrators, health IT leaders, and clinicians tired of being surprised by bounce-backs. #AI #HospitalReadmissions #MountSinai #PredictiveAnalytics #DeepLearning #HealthIT #ElectronicHealthRecords #VitalSigns #NursingNotes #LACEIndex #AUC #30DayReadmission #PopulationHealth #ClinicalDecisionSupport #HospitalOperations #Business #Technology #FexingoBusiness Keep every episode free: buymeacoffee.com/fexingo