It’s just as difficult to predict if your doctor is running on schedule as it is to draw your own blood. Fortunately, a team of researchers from Massachusetts General Hospital has figured out a way to eliminate those frustrating long waits by developing a machine learning technique designed to accurately predict radiology appointment wait times.
Long wait times are a serious source of contention for patients and physicians. In order to meet fee-for-service quotas, doctors must over-book their patient schedules so they can make sure they’re earning a consistent salary. However, this appointment booking system only causes waiting room stress and disrupts the patient flow.
“We noticed that most patients who were dissatisfied with the displayed waiting times were delayed for longer than predicted, so the need for more accurate models became imminent,” wrote the researchers in their study published in the Journal of the American College of Radiology. “We also wanted to predict not only waiting times for walk-in facilities, but also delays for the scheduled facilities.”
The researchers led by Catherine Curtis, MS, first considered a variety of factors that impact radiology appointment wait time, including “date and time, scheduling conflicts, patient flow, and number of exam rooms.” They then experimented with 10 different machine learning algorithms, including random forest, support vector machine, multivariate adaptive regression splines, nearest neighbor and linear regression, elastic net, and neural network. They found that elastic net was the best model for determining wait times or delay times, and neural network and k-th performed the worst.