Uroflowmetry is the one of the most commonly used non-invasive test for evaluating children with lower urinary tract symptoms. However, studies have highlighted a weak agreement among experts in interpreting uroflowmetry patterns. This study aims to assess the impact of machine learning models, which have become increasingly prevalent in medicine, on the interpretation of uroflowmetry patterns.

The study included uroflowmetry tests of children aged 4–17 years who were referred to our clinic with lower urinary tract symptoms. Uroflowmetry patterns were independently interpreted by three experts in pediatric urology. Discrepancies in interpretations were jointly re-evaluated by these three observers, and a consensus was reached. Voiding volume, voiding duration, and urine flow rates at 0.5-second intervals were converted into numerical data for analysis. Eighty percent of the dataset was used as training data for machine learning, while there maining 20% was reserved for testing. A total of five different machine learning models were employed for classification: Decision Tree, Random Forest, CatBoost, XGBoost, and LightGBM. The models that most accurately identified each uroflowmetry pattern were determined.

We included a total of 500 uroflowmetry tests in our study, comprising 221 boys (44.2%) and 279 girls (55.8%). The average age of the children was 9.17±3.41 years. In the initial assessment, 311 tests (62.2%) were interpreted identically by the observers, while 189 tests (37.8%) were interpreted differently by at least one observer (Fleiss' Kappa=0.608). Of the sample sused for machine learning training, 253 (50.6%) exhibited a bell-shaped pattern, 52 (10.4%) a tower pattern, 103 (20.6%) a staccato pattern, 40 (8%) an intermittent pattern, and 52 (10.4%) a plateau voiding pattern. Among the models tested, the highest accuracy was achieved with XGBoost (85.00%±2.90), while the lowest accuracy was observed with the Decision Tree model (81.80%±1.47). When evaluating voiding patterns individually, the intermittent voiding pattern demonstrated the highest accuracy rates (95%–100%), where as the tower (63.46%–73.08%) and plateau (61.54%–71.15%) patterns had the lowest accuracy rates.

The current trial demonstrated, for the first time, that machine learning models achieved a high accuracy rate in interpreting uroflowmetry patterns in children.

Consequently, artificial intelligence models have the potential to standardize the analysis of uroflowmetry voiding patterns in the future.