Optimisation methods in health care planning software

High demands on efficiency, resource utilization, service levels, and timeliness in the health sector lead to an increased need for optimizing complex decisions at strategic, tactical, and operational levels. Examples are nurse rostering and surgery sche duling. At the core of such decisions lie hard, discrete optimization problems. In operations of some size, human decision-making falls short of providing good solutions in time, and users need optimization based planning tools for support. There is a lar ge improvement potential that is chiefly determined by the quality of optimization methods. The optimization problems in question are particularly hard. Realistic response requirements typically exclude the use of exact methods. The only viable approach i s to use methods that produce high quality solutions in limited time. There is currently a strong development of such methods, based on so-called metaheuristics. Speed performance and adaptation to specific cases are two major, unresolved challenges. We s hall meet these by developing a methodology that exploits the emerging parallel processing capabilities of standard computers, synergies of multi-level search, and hyperheuristic strategies for self-adaptation. The SINTEF team consists of senior researche rs plus 1 PhD fellow. We shall collaborate closely with the CMA center of excellence at Uof Oslo. Uof Nottingham provides senior faculty, 1 postdoc, and 1 PhD. Two advanced Norwegian vendors, DIPS and Gatsoft, complete the consortium. In tight collaborati on, the team shall develop, exploit, and disseminate a methodology for solving hard optimization problems in health planning. For the two target applications we shall incrementally build prototypes for frequent industrial assessment. At least 25 papers wi ll be published in scientific journals and conferences. We shall disseminate results widely to end users and vendors.