Manuscript from my PhD work on making early prediction of osteoradionecrosis in head and neck cancer patients receiving radiation treatment has been published in Frontiers in Artificial Intelligence. Our manuscript is part of a special collection on ‘Artificial Intelligence in Precision Medicine’. Paper is available to read here.

Osteoradionecrosis (ORN), or bone death, is a major side-effect of radiation treatment in a specific type of head and neck cancer patients: oropharyngeal cancer. ORN is a late complication, developing around 2 years or more after radiation treatment is complete. In this paper, we show that capturing intricate changes in CT scans in the first 6 months after treatment can predict for the eventual onset of ORN with both high accuracy and precision. We used a combination of functional PCA and random forest to quantify these fine CT scan dynamics. This work hopefully can alert physicians early on to patients who have a high risk of ORN development enabling proactive therapeutic measures MUCH BEFORE the development of ORN!