Solving Long-Term Medical Trainee Scheduling Case study: Clinical Rotation Scheduling at Faculty of Medicine, Universitas Gadjah Mada

Medical training program is an important element of medical education. Besides of becoming one of the requirements for the students, it also becomes personnel fulfillment of some hospitals and clinics. Clinical rotation program is an example of medical trainee program. In the program, the trainees should perform some activities on some posts in certain periods. Scheduling the trainee to perform the activities needs to consider some rules and condition to make the program effective. This research attempts to develop a better model for medical trainee scheduling problem and to solve clinical rotation scheduling problem at Faculty of Medicine, UGM, Yogyakarta, Indonesia. Medical trainee scheduling is complicated due to the size and its constraints. For the clinical rotation scheduling, 11 constraints are exposed. All constraints are hard constraint except for maximum number of consecutive activities which is soft constraints. Soft constraint can be violated with some penalty cost while hard constraints cannot be violated. The objective function is to minimize total penalty cost. The problem can be formulated into three difference models i.e. the regular type model (Model 1), the proposed model (Model 2) and the modified model of Model 2 (Model 3). Model 1 uses assigning variables and setup variables while Model 2 only uses setup variable. Both models are applied to solve 8 problem instances using Branch & Bound method. To solve the problem, application programs are developed. The results show that Model 2 can represent the problem as well as Model 1 and it is better than Model 1 in term of solving time, total computation time, number of variables and constraints, and total nodes explored. However, Model 2 still cannot solve the real clinical rotation scheduling problem within 10 hours and therefore, it needs to be modified. The modified model (Model 3) uses artificial objective function instead of the real objective function and reduces the number of variables and constraints. The solving process is recursive. At the first trial, the problem is solved by setting the best possible objective function and finding a feasible solution. If a feasible solution cannot be found, the objective function value is decreased and a feasible solution is searched. The process will continue until a feasible solution is found. Using this approach, a solution that satisfies the users can be found in les then ten minutes. Compared with the manual scheduling, using the software for the scheduling will make the scheduling process faster and produce a better solution compared with manual scheduling. Moreover, the software can be applied to check some changes of conditions and rules and therefore, some preventive action can be taken to anticipate the problem that may arrise