#AI reads Urine# XGBoost-based urinary microbial signatures enable non-invasive diagnosis and prognosis for urothelial carcinoma
Published 07 December, 2025
This study focuses on urothelial carcinoma, a common malignant tumor of the urinary system characterized by high incidence and recurrence rates, where early diagnosis is crucial for treatment. The research enrolled 112 participants, including patients with bladder cancer, upper tract urothelial carcinoma, renal pelvis cancer, and healthy individuals. Midstream morning urine samples were collected from these participants, and 16S rDNA sequencing was conducted to analyze the diversity, community structure, and interactions of microorganisms in the urine. An XGBoost machine learning algorithm was used to build a diagnostic model, with data divided into 70% for training and 30% for testing, and external validation was also performed. Additionally, the SHAP algorithm was applied to interpret the key microbial features in the model. The study found that the diversity of microorganisms in the urine of patients with urothelial carcinoma was higher than that of healthy individuals, with consistent high levels of bacteria such as Streptococcus and Clostridium. There were significant differences in the microbial community structure among different cancer subtypes, and the microbial communities of healthy individuals and patients could be distinguished through analysis. The constructed diagnostic model showed good performance in detecting bladder cancer, with an AUC value of 0.927 in the training set and 0.811 in the external validation. Furthermore, some bacteria of the Lachnospiraceae family may be key biomarkers, which are not only associated with urothelial carcinoma but also may have prognostic value. This study proves that urinary microbial signatures can serve as non-invasive diagnostic and prognostic biomarkers for urothelial carcinoma. In the future, it is expected to develop a simple urine detection method based on this, facilitating the early detection, personalized treatment, and monitoring of urothelial carcinoma, and reducing the use of invasive examinations.
BMC Microbiol. 2025 Nov 12;25(1):732. doi: 10.1186/s12866-025-04455-w.
Youhe Gao
Statement: During the preparation of this work the author(s) used Doubao / AI reading for summarizing the content. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article.
For earlier AI Reads Urine articles：