AI-Based Prediction of Clinical Deterioration in Hospitalised Cardiac Patients.

INTRODUCTION: The rules based Between the Flags (BTF) system has been the mainstay of early warning of clinical deterioration in public hospitals in New South Wales, Australia. METHODS: A retrospective observational study compared BTF with an artificial intelligence (AI) model using gradient boosted trees with features augmented by time derivatives. RESULTS: The results showed that the AI model was able to provide several hours of additional warning time before adverse events. CONCLUSION: AI-based early warning systems may outperform standard care.