Prediction Modelling Presentation: Igor Shuryak and Everest Yang
Speaker Biographies
Igor Shuryak, MD, PhD is an Associate Professor at the Centre for Radiological Research, Department of Radiation Oncology, Columbia Irving Medical Centre. He focuses on quantitatively modelling radiation effects, including radiation carcinogenesis, tumour control and repopulation, normal tissue complications, radio resistance, and non-targeted effects. His research combines mechanistic radiobiological modelling with causal machine learning methods applied to radiation oncology and epidemiology.
Everest Yang is a Computer Science student at Brown University and the lead developer of the CAST framework. His work on time-varying treatment effect estimation using causal survival forests was presented at the three NeurIPS 2025 workshops (AI4Science, CauScien, and TS4H).
Abstract for the presentation
Predictive models in health research focus on answering “what will happen to this patient?” but clinicians also need the answer to a fundamentally different question: “what should we do for this patient?” The first is a prediction problem; the second is a causal one. In this talk, we present the CAST (Causal Analysis of Survival Trajectories) framework, a causal machine learning method that synthesizes horizon-specific treatment effect estimates from causal survival forests into smooth temporal curves, revealing when treatment effects emerge, peak, and attenuate. CAST enables both parametric (generalized least squares) and nonparametric (smoothing spline) trajectory fitting with uncertainty quantification. We apply CAST to two cancer settings: head and neck squamous cell carcinoma (presented at three NeurIPS 2025 workshops) and lower-grade gliomas using data from TCGA and CGGA (776 patients across six treatment-outcome scenarios). We discuss how causal survival methods complement rather than replace predictive survival models.
Please contact maudsley.brc@kcl.ac.uk if you have any questions.
Prediction Modelling Presentations
The Prediction Modelling Group at the NIHR Maudsley BRC hosts monthly online presentations on Machine Learning and Prediction Modelling and their applications to solve healthcare problems. Speakers from the UK and abroad present their works on developing and/or using machine learning and prediction modelling methods to answer questions as how to choose the best treatment for a patient or how to improve the diagnosis of a disease. Find out more: www.maudsleybrc.nihr.ac.uk/facilities/prediction-modelling-group/