Doctors and researchers are looking for better ways to manage neuroblastoma, a type of cancer that primarily affects children. A large review of 53 studies looked at how artificial intelligence, specifically machine learning models and hybrid nomograms, compares to traditional methods for diagnosis and predicting patient outcomes.
The data shows that these AI tools performed better than traditional markers when predicting a patient's prognosis. Specifically, AI-derived nomograms showed a higher area under the curve (AUC) of 0.9 compared to 0.8 for traditional gene signatures. While machine learning models also showed higher scores than radiologists in some areas, the difference was not statistically significant, meaning the results are still uncertain.
While these tools show potential, they are not ready for everyday use just yet. Many models were not properly calibrated or tested on outside data, and there is a lack of models specifically designed for children. Additionally, the AI models currently cannot accurately predict how a patient will respond to chemotherapy. These gaps mean that while the technology is promising, it still needs more testing before it can change how doctors treat patients.
