An artificial intelligence-powered learning health system to improve sepsis detection and quality of care: a before-and-after study

Comprehensive Summary

This study evaluated an AI-powered Sepsis Learning Health System implemented at Lausanne University Hospital to improve sepsis detection and care quality. The system integrates a standardized clinical pathway with an AI algorithm that retrospectively classifies patients as confirmed, possible, or invalidated sepsis cases every six hours. These predictions populate real time dashboards that track quality-of-care indicators and guide clinical interventions. Compared to control wards, hospital wards using SLHS showed reduced in-hospital and 90 day mortality for AI-flagged sepsis cases and increased accuracy of sepsis coding

Outcomes and Implications

This study demonstrates that clinician-integrated AI systems can meaningfully improve sepsis outcomes in real world hospital settings. By combining AI-driven surveillance with continuous quality improvement workflows, healthcare systems can enhance early sepsis recognition, reduce mortality, and improve data accuracy. The findings support broader adoption of learning health systems to strengthen public health responses to high-burden conditions like sepsis.