Health care systems are complex organizations with heavy demands on limited resources. For patients, this can lead to long waiting room times and other operational challenges that make a visit to the doctor a frustrating experience.
Researchers in the Johns Hopkins University Healthcare Scheduling Optimization Group are working to improve scheduling processes for both health care professionals and patients.
The group, part of the Malone Center for Engineering in Healthcare, focused one project on scheduling for rehabilitation inpatients—an area where the demand for therapists far exceeds the supply, and schedulers must manually sort through large amounts of information to match patients with the right therapist.
Anton Dahbura, MS ’82, PhD ’84, an associate research scientist in the Department of Computer Science, and students Anthony Karahalios ’17 and Quan Bui ’18 have developed a family of algorithms that can be used to assign therapists to patients based on a set of optimization criteria.
Krishnaj Gourab, assistant professor of physical medicine and rehabilitation, is testing how well the algorithm can prioritize actual patients at Johns Hopkins Bayview Medical Center. The algorithm is being tested in software that receives live data feeds from Epic, the medical records system used by Johns Hopkins Health System. “We have about 180 patients daily on the case load. Most have both occupational and physical therapy orders. So about 350 patients in total need to be seen by therapists in the hospital every day and will benefit from software that can streamline the scheduling process,” says Gourab.
Early testing shows the algorithm creates schedules that mimic the expertise of a human scheduler. “The schedulers are looking at the schedules created by this algorithm and saying, ‘Yes, this is exactly what I would have done,’” says Dahbura.
The next step: fine-tune the model, speeding up the scheduling process and allowing staff members to focus on patient care. “Right now it takes multiple dedicated staff members several hours to manage the process of assigning therapists to patients,” says Dahbura. “Our system does so instantaneously.”