Source/Disclosure
Shekhar S, et al. Summary #1413640. Venue: American Society of Clinical Endocrinology Annual Scientific and Clinical Conference. May 4-6, 2023. Seattle.
Disclosure: Shekhar does not report related financial disclosures. See research for relevant financial disclosures of all other authors.
key Takeaway :
- AI/machine learning showed high performance in detecting polycystic ovary syndrome.
- Using standard diagnostic criteria, AI may become a useful clinical tool for PCOS.
SEATTLE — Artificial intelligence and machine learning have shown high performance in diagnosing and classifying polycystic ovary syndrome, according to survey data.
With standardized diagnostic criteria, AI and machine learning could become more useful as tools to detect PCOS and reduce diagnostic delays, according to presenter at American Society of Clinical Endocrinology annual conference .
“The main takeaway from our study is the exceptionally high performance of AI and machine learning in detecting PCOS across different diagnostic and classification modalities.” Skand doctor shekhar A research assistant and endocrinologist at NIH’s National Institute of Environmental Health Sciences told Helio. “Specifically, studies that adopted standardized criteria for evaluating the performance of AI models in PCOS, such as the International PCOS Criteria and the Rotterdam Criteria, found that AI/machine learning I found that I was able to distinguish people who weren’t very good at it.”
In this systematic review and meta-analysis, Shekhar and colleagues searched Embase, the Cochrane Register, the Web of Science, and the Institute of Electrical and Electronics Engineers (IEEE) Xplore Digital Library to identify 31 observations from inception to January 2022. identified a study. AI/machine learning performance in detecting PCOS. Studies using clinical PCOS diagnostic criteria such as the NIH, Rotterdam, Revised International PCOS Classification were considered to diagnose PCOS, and studies without these criteria were considered to have classified his PCOS.
Studies included 9 to 2,000 participants. Overall, 23% of the studies were multicenter studies, mostly conducted in India (29%) or China (16%), and the median age of the PCOS participant he was 29 years. A total of 32% of studies diagnosed his PCOS using established diagnostic criteria, and the remaining 68% of studies classified his PCOS.
Clinical data with or without imaging were used in 55% of studies. The most common AI/machine learning techniques employed were Support Vector Machines (42%), K-Nearest Neighbors (26%), and Regression Models (23%). Researchers found areas under the receiver operating characteristic curve ranging from 73% to 100% in seven studies, diagnostic accuracy ranging from 89% to 100% in four studies, and ranging from 41% to 100% in ten studies. We observed a sensitivity of , and a specificity of 75%. Ten studies had a positive predictive value of up to 100%, four studies had a positive predictive value of 68% to 95%, and two studies had a negative predictive value of 94% to 99%.
“We were amazed at the tremendous effectiveness of AI and machine learning in detecting PCOS across a range of technologies and data. It promises to open up exciting new avenues to promote the health and well-being of the millions of women around the world who are underrepresented.
“Future research should employ AI in the setting of electronic health records and include input from more clinicians so that these technologies can be effectively adopted into the clinical care of patients with PCOS,” he said. said.
For more information:
Skand doctor shekhar You can contact us at skand.shekhar@nih.gov.
American Society of Clinical Endocrinology Annual Meeting
