| This research proposes a new architecture for knowledge based scheduling systems that operate in a dynamic environment. The study is based on an analysis of a real world scheduling problem, namely, the scheduling of catheterization laboratories (cath labs) for cardiac procedures in a large hospital. The hospital setting provided a particularly useful proving ground for research in scheduling and rescheduling due to the volatile environment and the critical nature of the events that occur in this environment. The problem is representative of domains where the scheduling process is complicated by a large number of potential disruptions that could affect the planned schedule.; A software engineering methodology was adopted for the research. A conceptual framework for the problem was developed by synthesizing our understanding from two perspectives: (a) an investigation of the scheduling processes and practices adopted by the expert schedulers in the real world setting using the protocol analysis technique, and (b) studying the relevant disciplines for approaches and ideas leading to system functionalities for supporting the problem requirements. Applying the prior Artificial Intelligence (AI) research in the areas of opportunistic planning and reason maintenance systems, a hybrid architecture that incorporates the twin features was proposed. Opportunistic plan strategies are used to incrementally construct the schedule, while the reason maintenance technique helps in schedule revision by undoing only parts of the schedule that are affected whenever changes take place.; Based on the architecture, and employing the data from the cath labs, a knowledge based system CLASS (Cath LAb Scheduling System) was developed and tested. Field testing of the system was carried out by having the expert scheduler at the cath labs evaluate the system functionality and output. The domain expert expressed strong agreement with the scheduling and rescheduling decisions of CLASS. The system was found to generate schedules consistent with the objectives and constraints of the domain, and to adjust well to the changing circumstances of the environment. |
