This article is more than 5 years old. Information may no longer be current.
Machine learning models may facilitate clinic workflow
Click Here to Manage Email Alerts
SAN DIEGO — Exam times predicted by machine may be used to help manage clinic workflow and patient wait times, according to a speaker.
“We know physicians are pressured to see more patients in less time, which can create patient wait time and clinic inefficiencies,” Michelle R. Hribar, PhD, of Oregon Health and Science University, said at the American Association for Pediatric Ophthalmology and Strabismus annual meeting.
Because patient wait times can be a driver of patient satisfaction, Hribar and colleagues developed a scheduling template based on patients’ anticipated exam length, categorized as short, medium and long, and based on visit type. For example, a new patient visit takes longer than a return visit, adult strabismus exams are long, and retinopathy of prematurity infant checks are short.
“We found that scheduling the short patients first and the long patients near the end of the clinic reduced average patient wait times,” Hribar said. These data are published in Ophthalmology.
To improve on that premise, the researchers focused on developing a machine learning model that could predict provider exam time, which is seen as the “bottleneck” in the clinic workflow, she said.
Input data were culled from 8,675 office visits for five pediatric ophthalmology providers over 1 year. Seventy-five percent of the data was used to train the model and 25% was used to test. Ten-fold cross-validation was used to increase the robustness of the data.
The most important features that contributed to accuracy of the prediction were “which provider the patient saw, whether or not their eyes were dilated, what their average prior exam length was, their age and their diagnosis,” Hribar said.
When the model was compared with provider predictions, the model was accurate about 65% of the time and the provider was accurate about 42% of the time.
“This shows some promise as to whether this model will work,” she said. – by Patricia Nale, ELS
References :
Hribar MR. Using machine learning to predict pediatric exam times. Presented at: American Association for Pediatric Ophthalmology and Strabismus annual meeting; March 28 to 31, 2019; San Diego.
Hribar MR, et al. Ophthalmology. 2019;doi:10.1016/j.ophtha.2018.10.009.
Disclosure: Hribar reports no relevant financial disclosures.