Radiotherapy is an essential part of treatment for most children with brain cancer. While it can be life-saving, it often causes long-term side effects. One of the most common and serious is permanent hearing loss, which affects around 60% of children survivors, detrimentally impacting their speech, learning, and quality of life. Children under 5 who receive high doses of radiation near the ear, or also receive certain chemotherapy drugs are at especially high risk.
Doctors currently lack reliable tools to predict which children are at high risk. As a result, it is difficult to adjust treatment plans to protect hearing without compromising the effectiveness of therapy. In this project, we aim to develop a risk scoring model to estimate the likelihood of hearing loss in children before treatment begins.
Using data from 3 clinical trials of proton radiation—a newer form of therapy—we will analyze how different treatment factors, such as radiation dose and chemotherapy, interact with patient characteristics like age and tumor location. The model will use machine learning but will be designed to give practical and usable tools for clinical teams. This tool could help personalize treatment planning, allowing doctors to lower the risk of hearing loss when possible. It could also guide follow-up care by identifying children who need early hearing support. Our goal is to reduce long-term harm from treatment and improve the quality of life for childhood cancer survivors.
