PROJECT TEAM
Mikkel Brabrand, MD, PhD, Odense University Hospital
Troels Martin Range, Associate professor, PhD
THE NEED
Acute patient arrivals vary greatly, making it difficult for hospitals to plan staffing and bed allocation. This unpredictability often leads to overstaffing or staff being called in on days off, causing stress, inefficiency, and increased costs. Without reliable data-driven forecasting, hospitals risk overcrowding, reduced patient safety, and poor resource utilization. A precise forecasting tool is needed to optimize workflows and enhance both staff well-being and patient outcomes.
THE SOLUTION
Praemostro uses a probabilistic convolutional neural network to forecast hourly patient attendance and hospital-wide bed occupancy. It integrates real-time data—such as local trends, weather, and infection patterns—to deliver accurate predictions. Already in daily use at an emergency department, it reduces guesswork, improves planning, and enables proactive responses to crowding. It significantly outperforms methods based on historical averages or manual estimates
Praemostro - Odense University Hospital
Call 5 - 2024 | Call 3 - 2022
500.000 DKK
Clinical Area
Emergency medicine / Hospital management
Technology
HealthTech
PROJECT SUMMARY
Praemostro is an AI-driven forecasting tool that predicts patient attendance and hospital crowding up to 12 hours in advance. It helps hospital departments optimize staffing and patient allocation, ensuring efficient use of healthcare resources and improved patient safety.
CLINICAL IMPACT
Praemostro enhances patient safety by preventing overcrowding and ensuring timely care through better staff allocation. It improves job satisfaction and reduces burnout by minimizing unnecessary call-ins and optimizing work schedules. Hospitals benefit from cost savings, streamlined operations, and improved patient flow. By forecasting bed needs across wards two days ahead, the solution enables proactive hospital capacity management, improving care quality and outcomes.