U.S. Nursing Homes (94-facility research cohort)

Nursing homes face challenges in real-time monitoring of resident condition changes, fall risk, medication safety, and functional decline. Staff often lack timely alerts to intervene before adverse events occur, contributing to preventable falls, functional deterioration, and inappropriate medication use.

An EHR system enhanced with AI/ML was deployed across 94 nursing homes. The system analyzed 150 daily clinical data elements per patient, generating real-time alerts to staff regarding changes in resident conditions, acuity levels, fall risk, and medication monitoring. A difference-in-differences study design compared outcomes against 124 control sites using standard EHR only.

Statistically greater improvements were observed in 16 of 18 CMS quality measures (89%) at EHR+AI sites. Major falls declined by 9%, residents requiring help with daily activities dropped by 22%, and the proportion of residents improving in functional status increased by 5%. Higher reductions in depressive symptoms and antipsychotic/antianxiety/hypnotic medication use were also noted, with stronger effects at higher-acuity and more diverse facilities.

• Real-time analysis of 150+ clinical data points per patient per day enables proactive staff intervention before adverse events occur.
• AI-enhanced EHRs show the strongest quality improvements at high-acuity facilities and those serving racially and ethnically diverse populations.
• Functional outcomes, fall prevention, and behavioral health metrics all improved simultaneously, suggesting broad systemic benefit rather than narrow optimization.