Adapting machine learning models to temporal drift: oral mucositis prediction in head and neck cancer patients receiving proton and carbon ion therapy

Key Points

• To improve the prediction of oral mucositis in head and neck cancer patients over time despite temporal drift.
• Utilized machine learning models that account for temporal drift.
• Employed transfer-learning ensembles to adapt models.
• Analyzed prediction accuracy based on therapies using proton and carbon ion.
• Assessed clinical relevance of toxicity predictions.
• Temporal drift led to significant reductions in prediction accuracy initially.
• Transfer-learning ensembles maintained reliable performance over time.
• Enhanced adaptability resulted in clinically relevant predictions for oral mucositis.