Combining artificial intelligence and microphones to diagnose IBS
By applying artificial intelligence to an acoustic belt, researchers have come up with a novel way to diagnose irritable bowel syndrome (IBS) by listening to a patient’s gut.
Speaking at the Digestive Disease Week (DDW) 2018, Barry Marshall, director of the Marshall Centre, University of Western Australia and lead researcher of the study said: “We used acoustic sensing technology that was originally created to tract the munching sounds of termites to see if we could detect problems in the human gut.”
Marshall noted that IBS is “an extremely common disorder that is notoriously difficult to diagnose.” He said: “We (the team) wanted to find a way to listen to the rumblings and grumblings of the gut to identify patterns that characterize chronic gut conditions like IBS.” To do so, they utilized acoustic sensing technology that was originally created to track the sounds of termite munching on wood to see if it could be applied to the human gut.
In the study, the researchers developed a basic prototype belt that uses machine learning techniques to identify complex features and patterns of the sounds collected from within the abdomen. Then, they recruited two groups of study participants--one with an existing clinical diagnosis of IBS and another with healthy digestive systems. Study participants were instructed to wear the belt and their bowel sounds were recorded for 2 hours post-fasting, and then for 40 minutes after a standardized meal.
Preliminary results from the study revealed that the acoustic index output of the belt predicted IBS with high accuracy, thereby allowing researchers to effectively differentiate between the two groups. Recordings from the first 31 of the IBS and 37 healthy participants were used to build the IBS acoustic index model. A statistical method called leave-one-out cross-validation was used
on this data set and yielded 90 percent sensitivity and 92 percent specificity for IBS diagnosis. Independent testing using the next 15 IBS and 15 healthy subjects revealed 87 percent sensitivity and 87 percent specificity for IBS diagnosis.
According to Joseph Muir, PhD, associate director of the Marshall Centre, University of Western Australia, the study allowed them to achieve proof of concept. Going forward, the team plans to further develop and test the belt on more patients, with the final aim of using it in a primary care setting for the diagnosis of IBS. “The hope is that this new technology can offer a less invasive way to diagnose this painful and sometimes debilitating condition.”
IBS is a common and often painful condition that causes bloating, diarrhea and constipation. It is estimated to affect more than 10 percent of the world's population. However, IBS can be difficult to diagnose and often requires patients to undergo a colonoscopy. Many patients with IBS remain undiagnosed and untreated. [Available at https://www.aboutibs.org/facts-about-ibs/statistics.html]