AI is now so smart that it can diagnose urinary diseases by listening to the sound of urine

An AI artificial intelligence algorithm can listen to a patient's peeing sounds, to effectively and successfully identify abnormal "flows" and corresponding health problems. This deep learning tool, called Audioflow, has so far worked almost like a dedicated machine in clinics, providing similar results to urologists.

Dr. Lee Han Jie from Singapore General Hospital, who led the research, said that there is a tendency to use machine learning technology in many areas because clinicians do not have much time to follow up closely. for each patient. In addition, especially since the pandemic, there has been a shift towards telemedicine and less direct hospital care. Their research team wanted to develop a way to monitor patients to assess their condition between visits to the hospital.

 This AI is even capable of outperforming some medical workers. (Illustration).

The algorithm for assessing the sounds produced by urine was performed in a soundproof environment, but the researchers hope to develop an application that is self-sufficient enough that patients can use it to self-monitoring at home. Current tumorimetry is effective in evaluating urological conditions but requires the patient to urinate into the machine during outpatient visits. 

The Covid-19 pandemic has limited access to clinics for patients. The team of experts wanted to develop a more efficient way to assess urine at home without any medical assistance and so they recruited the help of the engineering department to develop an algorithm. urine assessment. To train and validate this algorithm, they recruited 534 male participants from December 2017 to July 2019. The process was quite simple: the participants used a regular urine flow meter. in a soundproof room and record their urination with a smartphone. 

Using only about 220 recordings, the AI ​​learned to accurately assess flow rate, volume, and duration, all of which can indicate a blockage or problems with the bladder. This AI is even capable of outperforming some non-specialist medical staff and comes close to the level of higher-level consultants.

But the real benefit is that having a consultant in the bathroom equals you every time you pee. The team is currently working towards making the algorithm work in the presence of background noise in a normal home environment, which would turn out to be a different reality for patients.

The Audioflow instrument produced results that could rival a conventional urologist and a panel of six urology residents. The AI ​​technology reached a conclusion consistent with conventional urinary flow measurement for more than 80% of the recordings, and compared with urologists, it achieved an 84% agreement rate. The researchers now also hope that the new AI could soon prove beneficial in home settings.

Audioflow will be rolling out soon as a smartphone app for real-world testing with lots of background noise. For now, it still has the limitation that it can only be tested on different male urine stream sounds than female urine flow. A female-focused version may be ahead.