Predicting non-muscle invasive bladder cancer outcomes using artificial intelligence: a systematic review using
Jethro C C Kwong1,2, Jeremy Wu3, Shamir Malik3
1Division of Urology, Department of Surgery, University of Toronto, Toronto, ON, Canada.
NPJ Digital Medicine
|April 18, 2024
Summary
Artificial intelligence (AI) shows promise in predicting non-muscle invasive bladder cancer (NMIBC) recurrence and progression, but most current studies are low quality. Rigorous methodology is needed to ensure AI
Area of Science:
- Uro-oncology
- Artificial Intelligence in Medicine
- Clinical Decision Support Systems
Background:
- Accurate prediction of recurrence and progression in non-muscle invasive bladder cancer (NMIBC) is critical for patient management and clinical trial eligibility.
- The clinical readiness of artificial intelligence (AI) applications for NMIBC outcome prediction remains uncertain despite significant research interest.
Purpose of the Study:
- To systematically review and critically appraise AI studies predicting NMIBC recurrence or progression.
- To identify common methodological and reporting deficiencies in existing AI research for NMIBC.
Main Methods:
- A systematic literature search was conducted across major databases (MEDLINE, EMBASE, Web of Science, Scopus) up to February 5th, 2024.
- The APPRAISE-AI tool was utilized to assess the methodological and reporting quality of included AI studies.
- Performance metrics of AI models were compared against non-AI approaches within the included studies.
Main Results:
- Fifteen retrospective studies were included, focusing on NMIBC recurrence, progression, or both.
- The majority of studies were assessed as low quality, with only one deemed high quality.
- AI models generally demonstrated superior performance (accuracy, c-index, sensitivity, specificity) compared to non-AI methods, though the benefit varied with study quality.
Conclusions:
- Significant methodological and reporting pitfalls, including dataset limitations and heterogeneous outcome definitions, hinder the clinical readiness of AI in NMIBC.
- Collaborative efforts between urology and AI experts, coupled with rigorous research methodologies, are essential to develop high-quality AI models for improved NMIBC care.
