A 'smart toilet' in development uses artificial intelligence (AI) to scan stool for consistency and presence of blood — and early evidence suggests it is more accurate than patient self-reporting, a study reveals.
The remote, automated, real-time analysis and reporting increase the likelihood of physicians detecting gastrointestinal (GI) issues earlier, investigators report.
In a proof-of-concept study, the smart toilet was 85% accurate in categorizing stool consistency as loose, normal, or constipated. The findings were presented at the Digestive Disease Week (DDW) 2021 virtual meeting.
"This study highlights a very innovative and practical tool that could have major implications for patients and clinicians alike," Andrea Shin, MD, who was not affiliated with the research, told Medscape Medical News.
"Stool form or consistency and signs of bleeding are some of the most important pieces of clinical history when it comes to GI or bowel symptoms," added Shin, assistant professor of medicine in the Department of Gastroenterology & Hepatology at Indiana University School of Medicine in Indianapolis.
Image Analysis
The researchers tested their AI algorithm on 3328 images. They assessed photos from the internet and some submitted anonymously by participants in the study.
Two gastroenterologists also rated a subset of 552 images. The physicians showed "satisfactory agreement" on interrater reliability (the extent to which two or more 'raters' [eg, observers, examiners] agree), investigators noted.
