Head and neck cancer patients often face unpleasant side effects from radiotherapy, like dry mouth (xerostomia) and sticky saliva. Scientists are exploring ways to predict these side effects using machine learning. A recent study compared different machine learning models to see which one could predict these side effects the best.

The study included 85 patients who had radiation treatment for head and neck cancer. Researchers collected a lot of data from these patients, including information from CT and MRI scans, dosimetry (the amount of radiation received), and basic patient details. They used these data to build and test different machine learning models.

Eight different models were tested: eXtreme Gradient Boosting (XGBoost), Multilayer Perceptron (MLP), Support Vector Machines (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), Naive Bayes (NB), Logistic Regression (LR), and Decision Tree (DT).

The results showed that the Support Vector Machine (SVM) model did the best at predicting early and late side effects. For early sticky saliva and xerostomia, the SVM model had an AUC (Area Under Curve) of 0. 77 and 0. 81, respectively. For late side effects, the SVM and MLP models did quite well too, with an AUC of 0. 85 and 0. 64.

This study found that using a mix of data from CT and MRI scans, dosimetry, and patient details can help predict how badly radiotherapy side effects will hit a patient. Machine learning, especially the SVM model, can provide important insights to help doctors plan personalized treatments and reduce side effects for head and neck cancer patients.

How Machine Learning Can Help Predict Radiation Side Effects in Head and Neck Cancer

Actions