Using machine learning to predict suitability for surgery at an ASC vs HOPD

Editor's Note

This study from the University of Michigan, Ann Arbor, examines the use of a machine learning (ML) model to predict the suitability for having a surgical procedure performed at an ambulatory surgery center (ASC) vs a hospital-based outpatient department (HOPD).

To augment a labor intensive manual process in which physician assistants in the preoperative clinic review each patient’s suitability for having a surgical procedure at an ASC vs an HOPD, the researchers developed an ASC site optimization model that uses ML to stratify patients.

The ML model has two components: one predicting the outcome of clinician review for a patient’s suitability for surgery at an ASC and a second for whether the surgical procedure will be done at an ASC or an HOPD. Patients were stratified into six categories.

The models were trained using some 30,000 clinician reviews and surgical location types.

The researchers estimate that using the model-based reviews at their institution would save about 80 hours per week, which is equivalent to two full-time clinical employees.