General optimisation problems in health
Health care logistics: routing and scheduling
Many healthcare systems manage assets or workforces that they need to deploy geographically. One example, is a community nursing team. These are teams of highly skilled nurses that must visit patients in their own home. Another example, is patient transport services where a fleet of non-emergency ambulances pick up patients from their own home and transport them to outpatient appointments in a clinical setting. These problems are highly complex. For example, in the community nursing example, patients will have a variety of conditions, treatments may be time dependent (for example, insulin injections), nurses will have mixed skills and staffing will vary over time.
Routing and scheduling examples
Healthcare problem with multiple objectives
In health systems we are often trying to juggle multiple objectives. Emergency stroke care a classic case study. For example we might be minimising ambulance transport time to hospital, whilst trying to consolidate the stroke units in order to improve quality of care.
When considering optimisation of multiple objectives, the Pareto front is that collection of points where one objective cannot be improved without detriment to another objective. These points are also called ‘non-dominated’. In contrast, points not on the Pareto front, or ‘dominated’ points represents points where it is possible to improve one or more objectives without loss of performance of another objective.
Local Search Procedures
Local search procedures are classical metaheuristics that are used widely in industry. They are applicable to a variety of health service problems from assigning nurses to shifts, to routing and scheduling to ambulance location.