Artificial intelligence as a tool for adult men with urinary problems

Abstract

Background: Urinary problems are common among older men. In a Swedish study of men aged 40–80 years, the prevalence of lower urinary tract symptoms (LUTS) was 24%, but only 4% sought medical care. A Danish survey reported similar prevalence rates, with over 20% of men not discussing their symptoms with anyone. The incidence of LUTS is expected to rise due to an aging population. Artificial intelligence (AI) has been proven valuable in radiology, but its use in primary care remains limited. This study aims to develop an intelligent technique (IT) solution: a user-friendly mobile app that provides individualized support for men with LUTS. The secondary aim is to enhance diagnostic quality in primary care through AI-based decision support. Methods: A retrospective patient database was created, containing patient-reported symptom scores (IPSS), urinary diaries, and timed micturition assessments. These data were analyzed using machine learning and Bayesian methods to develop algorithms for personalized recommendations. A prospective clinical study, initiated in 2021, collects data from men seeking care for LUTS to expand the database. A randomized study will test the app's efficacy. Results: The first version of the app showed high concordance between its algorithmic assessments and urologist evaluations. A prospective study enrolled 50 patients from three primary health centers. One third of the patients did not get a recommendation at the first visit. Of the remaining less than half received the same treatment recommendation at this visit as the AI algorithm would have suggested. A web-based study is currently being initiated. Recruitment strategies include public campaigns and targeted invitations. Participants complete a digital questionnaire and are randomized if they meet the inclusion criteria. Conclusion: LUTS significantly impacts men's health, yet care-seeking remains limited. AI-based solutions, such as mobile apps for diagnosis and personalized recommendations, show promise as tools to improve healthcare for this population.