Machine-learning algorithms may be more accurate at setting lengths of time for OR procedures than humans. (Photo courtesy of Duke)
Researchers at Duke Health say that AI algorithms are more accurate than human schedulers at predicting the length of time needed to perform surgery, potentially reducing risk of errors and thousands in overtime costs.
In a study comparing their three AI models trained on thousands of surgical cases, the researchers found the algorithms were 13% more accurate in predicting lengths of time for 33,815 cases performed in outpatient and inpatient settings, and that reductions in overtime hours for cases that went past regular working hours resulted in about $79,000 in savings over four months.
“The model assisted schedulers to predict 3.4% more cases within 20% of the actual case length and 4.3% fewer under-predicted cases,” wrote the authors in their study.
The algorithms predicted 11.2% fewer under-predicted cases, and 5.9% more within 20% of the actual case length, and over-predicted 5.3%.
Duke University Hospital is currently using the solutions, and says the study is proof that machine learning models can be put into practice immediately in real-world scenarios.
“We created a unique framework that is being leveraged every day to predict surgical case length more accurately at case posting time and could be potentially utilized to deploy future machine learning models,” wrote the authors.
The findings were published in the Annals of Surgery.
Duke did not respond to HCB News' request for comment.
