A machine-learning approach to predict additional treatment after Bacillus Calmette-Guérin induction in non-muscle-invasive bladder cancer
A machine learning framework can be built to predict bladder cancer treatment failure, establishing feasibility though clinical usefulness remains unproven.
A multi-center Japanese real-world study (n=1524 NMIBC patients) demonstrates feasibility of ML modeling to predict BCG treatment failure, but class imbalance and limited performance signal indicate that clinically actionable prediction has not yet been achieved. The study provides an infrastructure proof-of-concept and identifies methodological requirements for future work.
What the study was
- Study design
- Retrospective real-world ML model development
- Population
- NMIBC patients from Japan Medical Data Vision database (April 2008-March 2024)
- Sample size
- 1524
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- Scientific Reports
Why it surfaced
Negative/inconclusive ML study for bladder cancer BCG outcome prediction — important for calibrating pipeline expectations. Large real-world Japanese dataset but model performance did not reach clinical utility threshold.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.