SHAP-based interpretable machine learning with longitudinal delta-radiomics across seven weeks of treatment for xerostomia prediction in head-and-neck cancer
Real-time imaging during radiation therapy predicts mouth dryness early enough to potentially adjust treatment and reduce lasting damage.
This study applied SHAP-based interpretable machine learning to longitudinal delta-radiomics features extracted weekly across 7 radiation treatment weeks to predict xerostomia in head-and-neck cancer patients, demonstrating improved predictive performance. Longitudinal on-treatment radiomics offers the potential for real-time adaptive treatment modification to reduce radiation-induced salivary gland toxicity.
What the study was
- Study design
- Retrospective longitudinal radiomics + ML prediction model
- Population
- Head-and-neck cancer patients undergoing radiotherapy; Malaysia
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- Radiation Oncology
Why it surfaced
Novel longitudinal delta-radiomics approach for treatment toxicity prediction with SHAP interpretability; addresses quality-of-life endpoint in HN cancer radiotherapy.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.