|Institution:||University of Minnesota|
|Keywords:||Lasso; Operative time prediction; Surgical team collaboration|
|Full text PDF:||http://hdl.handle.net/11299/162326|
Background: Operating room managers need to construct the surgery schedule for the next day by synthesizing information on estimated surgery duration, staff information, and surgeons' information. The purpose of this study is to assist operating room managers' decision making one day before the surgery by developing the predictive model for operative times taking into account the staff information. Methods: 10,960 cases in a health system in Middlewest are analyzed. The outcomes are the mean absolute errors of the predictions and the correlation between the predicted operative time and the observed durations, and the predictors include surgeon-scrub nurse pair IDs, individual surgeon IDs and individual scrub nurse IDs. Lasso regression modeling on the logarithm of the operative time is performed. Results: The unexplained variation of the residuals of the model, which only includes log scheduled duration and procedure type, can be further explained by the surgeon-scrub nurse collaboration frequency. Besides, the model that include surgeon-scrub nurse pair IDs, surgeon IDs and scrub nurse IDs can reduce the mean absolute errors by 8.47 minutes, compared with the scheduled procedure duration. Conclusion: The more surgeons and scrub nurses collaborate, the less time a surgery will take. Including surgeon-scrub nurse pairs in the predictive model, the prediction errors can be reduced.