Machine learning model significantly improves surgical time predictions

A study utilizing electronic health record data from 16,159 patients has demonstrated that an artificial neural network (ANN) model significantly outperforms traditional scheduling methods for predicting surgical procedure durations. The ANN achieved a root mean squared error of 49.7 minutes and a mean absolute error of 31.8 minutes, compared to an 18.52-minute discrepancy in surgeon estimates. By minimizing subjective bias in scheduling, this approach has the potential to optimize operating room resource utilization.