Opportunity

Patients, clinicians, and non-clinical staff depend on accurate operating room (OR) scheduling to ensure procedures start and end on time whenever possible and to standardize workforce operations. Current algorithms for predicting case length fail to account for known sources of variation, resulting in inaccurate time estimates. The downstream consequences result in prolonged patient wait times, unpredictable hours for clinical workforce, and elevated costs for the health system. 

Intervention

We partnered with the Perioperative Services team at the Hospital of the University of Pennsylvania (HUP) to implement the modeling tool ORACLE (Operating Room Accurate Case Length Estimation), which uses data analytics and machine learning to accurately forecast OR case times. ORACLE’s prediction algorithms draw on information from the electronic health record and professional claims to capture relevant information about patient multi-morbidities, clinician experience, and the scheduled procedure to more accurately estimate case duration.

Impact

Over one year of use at the HUP OR, ORACLE resulted in a 5 percent improvement in accurately scheduled cases, reducing total scheduling error by 884 hours. (A case is “accurately scheduled” if its actual length is within 30 minutes of its scheduled length.)  

We are currently exploring opportunities to expand use of ORACLE to other sites in the health system. Our long-term goal is to build a streamlined and assistive OR scheduling system incorporating ORACLE models. Implemented enterprise wide, this intervention could reduce OR idle time, thereby reducing the OR operating cost, and improve patient and clinician satisfaction by enabling more predictable schedules.