Case Story
Odense University Hospital

The software we offer has been in use for over a year in the emergency department at Odense University Hospital in Denmark.

Over time, staff have come to trust the predictions and incorporate them into their daily routines. And they can indeed trust the system. More than 95% of the time, it predicts with an margin of error below 5%.

Since August 2022, the Emergency Department at Odense University Hospital (OUH) has been using the system.

In this case story, you can read more about what one of the experienced nurses in the department has to say about it.

 

With our combined expertise in healthcare and mathematics, we developed a computer system that uses machine learning (AI) to predict activity levels up to 12 hours into the future. It draws on various data sources, such as typical workload patterns, major events, weather, and more, using this information to perform a series of complex mathematical calculations.

The result is a graph showing the expected number of patients hour by hour. We provide not only the predicted patient numbers for each hour but also the confidence level of the calculations.

This allows staff to determine how much weight to give the predictions and decide on the appropriate actions to take.

Praemostro has been in use in the emergency department since August 2022. Over time, the staff have come to trust the predictions and incorporate them into their daily routines—and with good reason. More than 95% of the time, the system is accurate within ±1 patient per hour, which is precise enough for any deviations to be easily managed within operations.

How the system also helps

Sick leaves

When there are sick leave reports, the system is used to determine whether the staff on duty can manage the expected number of patients or if additional personnel need to be called in.

Healthcare workers are often called in, so it’s preferable not to disrupt their days off unless absolutely necessary.

If the system indicates that it won’t be a busy day, the remaining staff can handle the tasks on their own, allowing the off-duty team to enjoy their well-deserved, undisturbed time off.

Planning

Although patients often arrive in patterns—most commonly in the late afternoon and early evening—the flow can vary significantly.

The staff use the system to plan their workday efficiently. It helps them decide when to take breaks, when they’ll have time to tidy up, and when they can eat.

The system provides valuable insights to help optimize these aspects of their day.

Workload

When the workload starts to increase, it’s crucial to stay ahead. If tasks exceed what the staff can manage, they risk falling behind. While healthcare professionals are used to working under pressure, there’s always a limit. If they are unable to keep up, it impacts patient care—hospital stays become longer, and the risk of complications increases.

This is why staff turn to our models when things intensify. If it’s a short-term spike in activity, they can manage it, and the backlog is quickly cleared. However, during prolonged periods of busyness, additional help is required. This could mean calling in staff from home or reallocating personnel from other parts of the hospital.