Researchers have built a machine that mimics bodily sounds to train AI — and it could be the future of disease outbreak detection in underserved communities
A team of scientists has built a machine designed to do something most people would never expect from cutting-edge medical technology: listen to the bathroom.
As explained here, the device, which the researchers have named the Synthetic Human Acoustic Reproduction Testing machine — or S.H.A.R.T. — is an engineering contraption equipped with pumps, nozzles, and tubes that mechanically recreate the sounds of human bodily functions. Far from a novelty, however, the project has serious medical ambitions: training artificial intelligence to detect disease outbreaks by analyzing the acoustics of urination, flatulence, and defecation.
The team presented their findings at the American Physical Society’s annual Fluid Dynamics conference, though their results are still awaiting peer review.
Listening where Doctors can’t
The core idea behind the project is surprisingly practical. Self-reported health data is notoriously unreliable — people forget, exaggerate, or simply don’t notice subtle changes in their bodily functions. The researchers, based at the Georgia Tech Research Institute (GTRI), wanted to find a completely passive, non-invasive alternative.
“Self-reporting is not very reliable,” said David Ancalle, a mechanical engineering student at Georgia Tech. “We’re trying to find a non-invasive way where people can get a notification on whether or not they should go get checked out.”
The system would work by pairing AI software with inexpensive acoustic sensors, quietly monitoring bathroom activity and flagging abnormalities. Think of it as a smoke detector — but for your gut health.
Building the machine
Getting the AI ready required real-world training data. The team sourced publicly available audio and video recordings of various bodily functions, analyzing the frequency spectrum of each type of event. Once trained, the AI needed a reliable way to be tested — which is where the SHART machine came in.
“A lot of thought went into each of the sounds,” said Maia Gatlin, an aerospace engineer at GTRI. “There was a subsystem for each sound on this little machine.”
By pumping water through the device in carefully calibrated ways, the machine reproduced a range of human outputs with enough fidelity to serve as a meaningful testing environment. The results were striking: the AI correctly identified the right type of “excretion event” with 98% accuracy in early testing.
“It actually performs pretty well,” Gatlin noted.
A tool for the world’s most vulnerable communities
The researchers aren’t building this for high-tech hospitals. Their focus is on regions with fragile health infrastructure, where disease outbreaks can spiral quickly due to delayed detection.
Cholera, for example, spreads rapidly through contaminated water and causes severe diarrhea — often in densely populated, under-resourced communities. At the time of the conference presentation, a cholera outbreak was already underway in Haiti.
“There’s an outbreak and resurgence in Haiti as we speak,” Gatlin said. Faster case detection, she added, could make a meaningful difference in containing such crises.
The vision is a networked system of cheap, AI-enabled sensors deployed across affected areas — quietly listening, aggregating data, and surfacing early warning signals.
“As we classify those events, we can start to collect that data,” Gatlin explained. “It can say, ‘Hey, we’re seeing an outbreak of lots of diarrhea.’ Then we can start to quickly diagnose what’s going on in an area.”
Affordable by design
Crucially, the team is committed to keeping costs low. The technology is specifically engineered to be accessible, not exclusive.
“We’re not trying to come up with million-dollar equipment,” Ancalle said. “We are trying to make this something that can be afforded by everyone — particularly since the project is focused on urban areas with weak health systems. The affordability aspect is very important for us.”
If the technology continues to develop as hoped, the humble sounds of the bathroom may one day become one of public health’s most powerful early warning systems.