Machine learning helps forecast and improve workloads for hospital cleaning staff
As cleaning staff at Danish hospitals continue to reduce in size, the need for effective utilisation and task prioritisation has never been greater. In the North Denmark Region, Aal-borg University Hospital now uses machine learning to pre-cisely forecast workloads and improve task management for cleaning staff.
For the past 10-15 years, cleaning staff across Danish hospital have gradually been reduced due to cutbacks and prioritisation. At the same time, cleaning and housekeeping services have remained a vital part of hospital operations. In the North Denmark Region, a machine learning algorithm embedded in Systematic’s Columna Flow Task Management now supports hospital cleaning staff by forecasting workloads to increase utilisation and prioritise the right tasks at the right time.
The intensity of cleaning and service tasks usually vary across weekday, weekends, and holidays, but also from department to department. Some tasks require significant effort while others are easier to finish quickly. By employing online learning, the task management solution constantly analyses and learns from task data to continuously evaluate progress in working groups and forecasts how their day will end: Either ‘Behind schedule’, ‘on schedule’, or ‘ahead of schedule’.
Improvements by the hour
When teams are forecast to end behind schedule, the solution automatically suggests reallocating team members who are ahead of schedule to support struggling teams. This not only helps teams complete their tasks within their deadlines, but also alleviates over-worked teams. Currently, the solution offers forecasting with very high accuracy. The result is a more optimised deployment of cleaning staff, accurate staffing forecasts, and improved task management.
The vast potential in task data
As task data is generally consistent, task data has the potential to be used to forecast many different events e.g., patient arrivals, delays in consultations, and other potential bottlenecks. Together, these insights can be combined to unlock further machine learning-based improvements and ensure optimal utilisation of already scarce healthcare resources.
