Artificial Intelligence and Machine Learning
AI and ML enable us to learn from complex, high-dimensional cancer data that traditional methods cannot easily capture—especially when outcomes evolve over time and are influenced by multiple interacting factors. Our models integrate diverse data sources (clinical characteristics, PROs, genomics, labs, and treatment details) to predict adverse events and quality-of-life outcomes early in therapy. We focus on building models that are accurate, clinically interpretable, and designed for real-world deployment.
A major emphasis of our work is explainable prediction: identifying which symptoms, clinical features, or biological variables most strongly contribute to risk and how these drivers change over time. We also prioritize robust evaluation, fairness, and generalizability across patient subgroups. By translating AI insights into usable risk estimates and decision-support tools, our goal is to help oncology teams intervene earlier, personalize supportive care, and reduce the long-term burden of cancer therapy. We also employ deep learning models on pathology and radiology to predict clinical outcomes.