By Mark Collingwood, VP EMEA, FICO describes how digital aviation software incorporating FICO optimization tools helped organize nurse scheduling for Stockholm Karolinska University Hospital intensive care unit at the start of the COVID-19 pandemic

The arrival of COVID-19 turned everything on its head, creating an immediate need to rapidly expand intensive care units, and challenging existing employment practices and regulation. Out of this crisis inspirational and clever collaborations around the world delivered a myriad of life-saving solutions. One example was the use of Boeing’s market leading Jeppesen digital aviation software, which uses FICO® Xpress Optimization, to solve a crucial nurse scheduling problem at the intensive care unit (ICU) of Karolinska University Hospital, Sweden’s second largest hospital.

The virus was spreading throughout Sweden, and Stockholm’s Karolinska University Hospital needed to scale up its operation significantly to cope with the rapidly increasing volume of patients. The situation was overwhelming and unprecedented. The staff were working at extreme levels, exposing and risking their health and life, but diligently coping with the influx of patients.

A larger workforce was established by adding people from various other care units and a Swedish Crisis Agreement, which changed the regulation around working hours and salaries, was put into effect. But Karolinska needed to create a plan to ensure that this much larger workforce, with a new work agreement, would operate safely and effectively. And the plan needed to be in place within a week to be able to produce schedules for the ICU’s nurses and assistant nurses.

The solution came from an unexpected source. The hospital approached Boeing‘s crew solution team in Gothenburg, to help them solve this urgent problem. Expanding the ICU and quickly implementing Boeing’s Jeppesen-optimized scheduling system enabled for more efficient, safer work and eased some of the burden the hospital faced through the height of the pandemic.

Against the clock

“The biggest challenge, by far, was time,” explained Daniel Roth, Senior Business Advisor with Boeing. “We only had a week to produce the initial schedule, which had to incorporate who could work when, individual nurse competences, special requirements with respect to their schedules, and other factors. This data was not available in a structured way, but rather in the heads of current schedulers and management. Fortunately, our extensively used aviation solution with FICO Xpress Optimization as an integral part, enables an end-user to quickly build schedules.”

Another challenge was accommodating the labor rules of the Crisis Agreement; it very quickly became apparent that the Crisis Agreement work limits exceeded those at which a human being can work for an extended period of time. Therefore, it was essential to establish reasonable workloads and shift patterns.

The resultant structure was based on two types of 12.5 hour overlapping shifts, meaning 56 per week. Seven different competence profiles were established, with minimum requirements per shift and an objective to maximize coverage. And, using Boeing’s Jeppesen Crew Rostering, enabled the creation of rosters for over 300 nurses and healthcare workers during the peak period, resulting in more workable shifts for staff and better coverage for the hospital.

Turning theory into practice

The optimization problems faced by the Karolinska Hopsital fall into the class of NP-Hard in complexity theory, meaning one cannot expect to solve larger instances exactly in reasonable time. Tomas Gustafsson, Portfolio Manager with Boeing explained how the team tackled this. “The Jeppesen solution approach combines heuristics and exact methods to reach those practical run times, where the sub-problem linear and integer programs are solved using FICO Xpress.”

Once a first solution was produced, a joint Karolinska-Boeing solution team intensively refined the schedule over of a few days and nights. This included determining what work patterns were possible, changing the problem slightly, re-running and producing a new solution. “None of this would have been possible without the speed, flexibility and robustness of the underlying optimization algorithm, of which FICO Xpress is an essential part,” added Gustafsson.

As such, it is a fantastic use of optimization to solve one of the most pressing problems worldwide last year. Indeed, anyone looking for examples of the positive benefits of using optimization should look at what Boeing and Karolinska did in a matter of days to help healthcare professionals manage the coronavirus crisis.

For its achievements, Boeing’s crew solution team won the 2021 FICO® Decisions Award for AI, Machine Learning & Optimization. “Boeing adapted its Jeppesen airline and railway scheduling optimization system to solve ICU nurse scheduling problems during the pandemic,” said Lisa Morgan, contributing writer at InformationWeek and one of the FICO Decisions Awards judges. “The judges were impressed how quickly Boeing’s team understood the problem, addressed it, and implemented a solution. These data scientists did an excellent job of rising to the occasion.”

More information: https://www.fico.com/en/products/fico-xpress-solver