Using Machine Learning in Diagnosis and Early Detection of Lower Urinary Tract Symptoms (LUTs)

Abstract

LUTs are not bothersome in the early stages so the majority of people with them ignore them until they are at severe stage and their quality of life is affected greatly, it’s at the severe stage that people are diagnosed and have to be admitted for monitoring at the hospital. The project is aimed to create a web application that will process the data obtained by the smart watch which the patient will be wearing monitoring their voiding events. We conducted a survey at Mulago referral hospital where we interview 6 patients, we used questionnaires to understand the what kind of LUTs they suffered from, when they noticed the first symptoms and how they are voiding events are affected by the LUTs. Basing on the information we focused two kinds of symptoms one about the frequency of urination and the other about the urine flow strength. We then built a machine learning model to identify the urine flow stream strength and built another algorithm on the web application to determine the frequency of urination. From the questionnaires all 7 patients suffered from obstruction LUTs and were above 50years, 1 patient urinated over 15 times a day and another frequency depended on the amount of water he took, all the patients confessed to not taking the problem serious in the early stages even with symptoms. In conclusion the patients admitted had remained in the hospital for more than 2 years not working at all, that showed that LUTs have to be treated in the early stages so that the bread winners of the families can continue to provide for their families.