Outcome prediction in head and neck cancer patients using machine learning methods

Head and neck cancers represent approximately 4% of new cancer cases in the United States in 2024, with an estimated 70,230 diagnoses and 16,100 deaths, highlighting their ongoing global health challenge. This study employs machine learning to predict clinical outcomes for 100 patients with head and neck cancer, using data from the HNSCC Collection in The Cancer Imaging Archive (TCIA). Six key features—cancer site, primary tumor volume, race, smoking status, HPV status, and treatment type—were selected to enhance prediction performance. The random forest model achieved an accuracy of 94.68% and an AUC of 0.9821, significantly surpassing the baseline majority classifier (50%). These results demonstrate the potential of this approach to support personalized treatment planning and improve clinical decision-making.