Deep learning approaches for predicting response to neoadjuvant chemotherapy in muscle invasive bladder cancer: A systematic review.
Mohammad Reza Nowroozi, Mohammad Hamidi Madani, Amirreza Farzin et al.
✦ AI-curated · Sources linked
In 30 seconds
This systematic review evaluated deep learning models for predicting neoadjuvant chemotherapy response in muscle-invasive bladder cancer (MIBC). Analyzing 12 studies with over 1,575 patients, the review found that area under the receiver operating characteristic curves ranged from 0.69 to 0.89, with hybrid models outperforming standard convolutional neural networks. Methodological quality varied, impacting clinical applicability.
Key findings
- Area under the receiver operating characteristic curves ranged from 0.69 to 0.89.
- Hybrid models consistently outperformed standard convolutional neural networks.
- Most studies employed convolutional neural networks, integrating various data types.
- Concerns regarding reproducibility and external validation were frequently noted.
Why it matters
Improving prediction of treatment response in MIBC can enhance personalized treatment strategies, potentially leading to better patient outcomes. Understanding the capabilities and limitations of deep learning models is crucial for their integration into clinical practice.
What to keep in mind
Methodological inconsistencies and limited external validation were noted, which may affect the reliability of the findings.