AI-powered smartphone app monitors IBS using poo images


Researchers in the US have developed an app that uses a smartphone camera and artificial intelligence algorithms to assess images of patients' poo for signs of disorders like irritable bowel syndrome (IBS).

The tool is intended as an alternative to patients' self-reporting stool form and frequency using the seven-point Bristol Stool Scale (BSS), which ranks consistency from hard to liquid but can produce highly variable results.

It is well recognised that the subjectivity of the BSS can make it challenging to get an accurate picture of therapeutic interventions designed to improve symptoms in gastrointestinal disorders like IBS in clinical trials.

In a 39-subject study reported in the American Journal of Gastroentereology, researchers at Cedars-Sinai Medical Centre in the US asked patients with diarrhoea-predominant IBS to take a picture of each of their stools over a two-week period using the app.

Those images were assessed by the AI for five visual characteristics – the BSS, consistency, fragmentation, edge fuzziness and volume – with the results double-checked by two specialist gastroenterologists. Another element of trial compared the AO to patient-reported BSS scores.

Overall, there was "good agreement" between the AI results and the experts on all five measures, and the app outperformed the self-reporting using the BSS system, with sensitivity and specificity 16% and 11% higher, respectively, resulting in fewer false-positive and false-negative results.

The authors note that studying GI disorders can be challenging,  particularly those that are deemed 'functional'  and rely on patients self-reporting symptoms like abdominal pain, bloating, and stool form and frequency.

There has been an evolution in the measures used in clinical trials of new therapies for IBS, from a weekly assessment of symptoms to daily self-reporting using the BSS.

However, the BSS "although intuitive, requires subject education and instruction to avoid confusion," they write in the paper.

"In this study, we show that AI is able to catalogue and characterise stool form in an objective way," the authors continue.

"Although the app is still dependent on having a subject to photodocument each bowel movement, the photos document the 'true' stool characteristics. This allows an objective measurement of stool form [and] allow the stool to be assessed with additional characteristics beyond the BSS."