In a recent study published in the American Journal of Gastroenterology, researchers at Cedars-Sinai Medical Center in the United States evaluated an application (application) for smartphones based on artificial intelligence (AI) trained to evaluate the characteristics of a patient’s feces.
Study: A smartphone app that uses artificial intelligence is superior to the subject’s self-report when assessing the shape of feces. Image credit: Josep Suria / Shutterstock
Fund
Functional gastrointestinal (GI) disorders, especially luminal ones, require the patient to self-report the shape and frequency of feces. However, because the symptoms of diarrhea common in patients with irritable bowel syndrome with diarrhea (IBS-D) are subjective, the inability to accurately report or assess the form and frequency of feces makes it difficult to determine. the effectiveness of therapeutic interventions in these conditions.
The Bristol Stool Scale (BSS) is a 7-point scale approved by the U.S. Food and Drug Administration (US-FDA) that rates stool consistency from 1 (hard lumps) to 7 (liquid). . However, inconsistent and inaccurate self-reporting of fecal forms causes problems, especially among IBS-D cases. In these cases, AI algorithms could help to systematically evaluate digital images of a person’s bowel movements.
About the study
In the present study, researchers enrolled subjects participating in a randomized clinical trial for IBS-D to validate AI determinations for stool images based on five different visual characteristics of feces, i.e., blurring of stools. edge, consistency, BSS, volume and fragmentation. In another set of individuals from the same trial, they assessed how the application findings were aligned with self-reported BSS scores. Finally, the team compared the stool characteristics scores determined by the subject and the AI with the standardized diarrhea severity scores.
Participating subjects captured all stool images during the two-week screening phase of the trial. The application processed the results and determined five visual characteristics of stool and bowel frequency. Two experts validated AI images from the first third of the subjects. Later, the team also classified the stool images annotated by AI, self-reported by study participants, and two experts, into categories, BSS <3 (estrenyiment), BSS ≥ 3, però BSS ≤5 (normal) i BSS> 5 (diarrhea). Finally, the team calculated the sensitivity, specificity, accuracy, and diagnostic probabilities of self-reported, AI-rated BSS scores, comparing them with expert evaluations, which considered the gold standard.
Study results
There were a total of 39 study participants, of whom 14 provided 219 stool images for the validation phase. The team used data from the other 25 subjects for the implementation phase. Both AI and expert gastroenterologists presented BSS scores of one to seven and their assessments agreed on the five characteristics of feces, as well as AI and expert assessments.
The mean rates of specificity and sensitivity of the categorization of the AI-qualified BSS score were 11% and 16% higher, respectively. The average diagnostic probability ratio and accuracy rate were higher for AI at 30.64 versus 3.67 and 95% versus 89% compared to the scores reported by the subject. The agreement between the BSS scores reported by the subject and those graded by AI was 0.31 during the validation phase, but reached a value of 0.61 during the implementation phase. On average, the visual characteristics of faeces determined by AI between the two phases remained similar.
In addition, the authors observed a good correlation between the mean daily BSS scores classified by AI and the severity scores of diarrhea in IBS-D subjects. The other four visual characteristics of feces reported by the application also correlate quite well with the severity scores of diarrhea. It should be noted that all subjects found the app easy to use, and 50% of those who responded to queries about the user experience described their experience as easy and very enjoyable.
Conclusions
Previously, IBS drug testing often relied on weekly assessments of gastrointestinal symptoms. Later, the U.S. FDA developed new guidelines for IBS, which required trial sponsors to ask all participants to report and characterize symptoms daily to improve accuracy. However, BSS remains critical to assess self-reported stool hardness and to assess individual stool types during clinical trials.
An inaccurate self-report of the stool form could result from an inadequate understanding of the subject and a memory bias. Although intuitive, the patient should be familiar with BSS to avoid misperception. It becomes a challenge when subjects with diarrhea report a daily average of BSS while having several varied bowel movements in one day. The findings of the current study support that self-reported daily scores differ from BSS scores given by both experts.
AI catalogs characterized the form of feces in a “true” objective sense, since a subject photographically documented each bowel movement. Digital stool images allowed a complete assessment of the effect of a drug and objectively quantified the side effects of therapies for intestinal disorders. In addition, these images assessed the characteristics of feces beyond the BSS. The evaluation of four new features facilitated the consideration of each bowel movement separately and avoided the need to compare daily averages. Overall, the observed pattern of test characteristics suggested that the AI results were superior. In addition, they reduced trial costs, as sponsors could now design trials with a large number of subjects to reduce the effect of self-report inconsistency and inaccuracy. Consequently, in the future, objective measurement of stool form would need fewer subjects to test drugs.
In conclusion, the AI-based application used in the study accurately characterized feces compared to self-report and correlated well with the severity of diarrhea. It has the potential to become a valuable tool for use in trials of luminal gastrointestinal diseases, including IBS-D, as it was accurate and objective in defining faecal characteristics beyond BSS. .
Magazine reference:
- Pimentel, Mark, Mathur Ruchi, Wang Jiajing, Chang Christine, Hosseini Ava, Fiorentino Alyson, Rashid Mohamad, Pichetshote Nipaporn, Basseri Benjamin, Treyzon Leo, Chang Bianca, Leite Gabriela, Morales Walter, Weitsman Stacy, Kraus Asaf, Rezaie Ali, A The smartphone app that uses artificial intelligence is superior to the subject’s self-report when assessing the shape of the stool, The American Journal of Gastroenterology: July 2022, DOI: 10.14309 / ajg.0000000000001723 ,