Knowledge-based schedule formulation and maintenance under uncertainty.

This paper is concerned with the dual sequential problems of (1) determining an acceptable personnel schedule over a specified time period, and (2) adjusting that schedule during the course of its execution in reaction to daily changes in both demand and available personnel. The first problem is schedule formulation; the second sequential problem is schedule execution. A rule-based, hierarchical system has been developed for first modeling and then solving both the schedule formulation and the schedule execution problems as a two-phase dependent process. The system is applied to the scheduling and staffing of nurses. A double-blind evaluation was conducted, which ascertained the quality of the resultant schedules in terms of maintainability, coverage, and personal satisfaction. The evaluation indicates that for units on which personnel changes have occurred, the prototype appears to perform as well as human schedulers.