Optimizing operating room planning

A great part of the population that is operated has to wait a long time to access the surgical procedure, and as time elapses the condition of these patients degenerates. On the other hand, the amount of time during which operating rooms are idle is signi cant. Thus, optimizing the operating theater becomes important to reduce the time patients wait for their treatment and to avoid unnecessary waste of resources. This dissertation presents a combination of two decision management techniques applied to the operating theater, in order to improve the e ciency of the surgery scheduling problem present in healthcare institutions. The approach developed integrates a data mining model, used to predict the duration of surgeries, and an optimization model, to handle the decision process of scheduling surgeries. The problem primarily exists because surgeries are naturally uncertain, resulting in high variance in the duration of surgical procedures. The inherent uncertainty of surgeries leads to deviations from surgeon duration estimates, disrupting schedules or causing under-utilization of operating rooms. To avoid this problem, a combination of supervised learning techniques are used to improve the accuracy of these estimates. Surgery scheduling is also a di cult combinatorial problem, in which surgeons have to nd a time and location to operate their patients. This problem has a strong impact in the performance of healthcare organizations and societies, hence the importance to optimize it. This approach is based in a mixed integer programming model, solved using exact methods and a meta-heuristic, developed for this purpose. This work shows that there is an opportunity to improve the performance of the operating theater with a generalized scheduling model. On the case study considered, the method proposed beats the surgeons own surgery duration estimates and outperforms their scheduling plans. The data mining methodology developed was tested in ten di erent surgical specialties and is able to increase the accuracy of surgeon estimates by up to 44%. The optimization component was applied to an outpatient surgery department, duplicating the number of surgeries performed and increasing operating room utilization by at least 50% during the period of time tested.