Event
PHUSE EU Connect 2023
05.11.2023
About the event
PHUSE EU Connect 2023 provided a platform for professionals from life sciences, clinical research, and data analytics to exchange ideas on current developments, emerging technologies, and regulatory challenges in the pharmaceutical industry.
HMS was also there, sharing insights and experience from our project work. The event offered plenty of opportunities for professional exchange, new perspectives, and discussions on how innovative technologies are transforming the way data is used in the pharmaceutical industry.
Our contribution to PHUSE
Talk
Machine-Learning Approaches to Survival Analysis of Clinical Data
Dr. Dorothee Childs, Christoph Bergen
This talk explores machine learning methods for survival and time-to-event analysis of clinical data. Starting with the classic Cox proportional hazards model, it compares established methods with newer machine learning approaches, including random survival forests, survival support vector machines, and boosted trees.
Using the Python package scikit-survival, the talk demonstrates how these models can be combined with model selection, cross-validation, and performance metrics. It also explores the use of SHAP to improve the interpretability of more complex models – a particularly important aspect in clinical applications.