Title: sparsesurv: a Python package for fitting sparse survival models via knowledge distillation

Abstract: In this talk, I will present our recent work on developing a software package to fit sparse semiparametric survival models via knowledge distillation (Wissel, David, et al. "sparsesurv: a Python package for fitting sparse survival models via knowledge distillation." Bioinformatics (2024)), also deemed preconditioning or reference models in statistics (Pavone et al. (2022), Paul et al. (2008)). After a general introduction to sparse semiparametric right-censored time-to-event models, I will discuss the advantages of choosing the scikit-learn API for our implementation, followed by an overview of results and a short demo. If time permits, I will also discuss recent work on sparse partially linear additive Cox models and benchmarking of multi-omics cancer survival models (Wissel, David, et al. "SurvBoard: standardized benchmarking for multi-omics cancer survival models." bioRxiv (2024)).

Biography: David Wissel is a fourth-year PhD student shared between ETH Zurich and the University of Zurich. He is primarily interested in applications of right-censored time-to-event models on cancer data, particularly in terms of interpretability and the use of multi-omics data. He also works on long-read RNA-seq data, primarily assessing its quantification and isoform discovery capabilities.

Join using this Teams link.

Please contact maudsley.brc@kcl.ac.uk if you have any questions.