The AI algorithm in an app makes a urological diagnosis based on the sound of peeing.
Artificial intelligence (AI) algorithms have been developed by researchers to recognise urine problems based on the sound of urination. The algorithm included in the smartphone app may turn out to be a practical and affordable tool for monitoring urological patients at home in the future, according to the creators, who spoke in Amsterdam at the European Urology Society’s annual congress.
In testing the parameters of urine, the deep learning system called Audioflow performed almost as well as the specialist equipment used in clinics and produced findings that were comparable to those of urology residents. The goal in the long run is to develop an app that patients may use to keep track of themselves in their typical home setting. The current study shows the first trial version of the programme, which analyses the noises associated with urine under soundproof conditions.
According to estimates, 57 percent of women and 60 percent of men experience lower urinary tract symptoms that are linked to functional issues with the bladder and urethra.
Uroflowmetry, which measures the characteristics of urine, is an essential tool for evaluating the state of symptomatic patients, but patients now have to urinate in a particular machine under clinic circumstances for the assessment. Patients are instructed to pee into a funnel that is attached to the uroflowmeter during the test, and the device records the characteristics of fluid flow as it happens. However, patients found it difficult or impossible to get to appointments during the COVID-19 outbreak, and even when they did, they had to wait in huge queues for the sole machine.
Dr. Lee Han Jie and his associates from the Singapore General Hospital created the learning algorithm in collaboration with experts from the engineering department. Between December 2017 and July 2019, 534 male participants who used the conventional uroflowmeter in a sound-proof room and recorded the noises with a smartphone were recruited to train and validate the software.
The sophisticated system learns to predict the flow rate, volume, and duration of urine based on 220 audio recordings. These measures can show…