An innovation platform sponsored by the Novo Nordisk Foundation
Praemostro.com – Odense University Hospital


Emergency departments are one of the most unpredictable areas of hospital operations, and hospitals understandably struggle to avoid being under- or over-staffed. Statistically, most patients arrive on Mondays and Fridays after lunch – but the full picture is a bit more nuanced and difficult to predict.

The Inspiration Behind the Innovation

Currently, there are no data-driven tools to assist hospital management with staffing decisions. Therefore, staffing decisions are largely guesstimates based on intuition and a loose reliance on past patterns. Data shows that this method will significantly underestimate demand int he future, leading to chronic understaffing – or overstaffing due to overcorrection. It’s a delicate balance. Understaffing causes workers to be called in on their days off, negatively affecting job satisfaction and work/life balance; while overstaffing leads to unnecessary expenses.

The Innovation 

Praemostro.com improves hospital management’s ability to schedule staff according to need. This substantially improves the likelihood that the necessary resources will be available to handle incoming patients while judiciously reducing unnecessary costs. The result is an adequately staffed department where patients receive a high standard of care and where clinicians aren’t overwhelmed by the number of patients they must care for. 

Praemostro.com is a machine-learning-enabled IT system that can reliably forecast demand in high-variability settings such as hospital emergency departments. Using datasets, the system can broadly predict demand months in advance. Even more impressive is that it has been shown to anticipate, with +/-1 accuracy, the influx of patients expected to arrive each hour over the next twelve hours.

The system's precision improves over time because it continually collects and uses calendar information, weather data, seasonal infection rates and other data to influence its predictions. The resulting predictive analysis can inform preliminary schedules months in advance with better accuracy than is possible today while facilitating optimal short-term staffing decisions for the next twelve hours.

The Team

Mikkel Brabrand: Consultant, Clinical Professor, PhD; Department of Emergency Medicine; Odense University Hospital

Troels Martin Range: Mathematician-Economist, Associate Professor, PhD; Department of Emergency Medicine; Odense University Hospital