A multi-agent system handling personnel shortages in hospitals

Within the field of nurse rostering[2], solving personnel shortages is a central problem. Trivedi and Warner[11] and Gascon et al. [5] introduce a pool of float nurses. They solve shortages by using nurses from this pool. Siferd and Benton[10] tackle shortages by calling nurses from other wards and by allowing nurses to work overtime. Meisels and Kaplanksi[6] pioneered a negotiation based approach. The latter authors introduce three stages. In stage 1, scheduling agents generate a local solution. In stage 2, a global feasible solution is constructed. The last stage is used by the agents to improve their local schedules. Meisels and Kaplanski assume a fixed pool of shared resources. Bard and Purnomo[1] propose various ways of tackling staff shortages. Relevant for the present discussion is the one in which initially negotiation takes place with a pool of float nurses. Consequently, nurses who are not needed by the wards, are placed on an on-call list. If none of the proposed solutions suffices the hospital can call in agency nurses. Among Belgian Hospitals the idea of a mobile equipe has been introduced, similar to the pool of float nurses described above. This equipe, modeled as a virtual ward, contains nurses that are assigned to wards where shortages arise. There is a mobile team per group of wards of a department that require employees with more or less similar skills. The wards managers are supposed to negotiate the assignment of the float nurses to the actual wards, taking the personal preferences of the float nurses into account. The manager of the department assigns the nurses when the managers of a group cannot reach agreement and is responsible for rosters of the nurses of the mobile equipe. The schedule for the float nurses is known one day before and may only be changed