Dynamic Prediction with Numerous Longitudinal Covariates - Mirko Signorelli

Описание: To make informed decisions, clinicians and patients rely on accurate predictions of the probability to experience adverse events such as dementia, cancer or death. Dynamic prediction models can update the probability of experiencing an event as more longitudinal data is collected. However, traditional joint modelling is computationally unfeasible with more than a handful of longitudinal covariates, and until recently R lacked a package that could deal with numerous longitudinal covariates. The R package pencal uses a penalized regression calibration approach that allows to overcome this limitation. It employs mixed-effects models to summarize the evolution of the longitudinal covariates, and a penalized Cox model to predict survival. Besides covering estimation, the package comprises functions to compute predicted survival probabilities for new subjects, and to validate model performance. For large datasets, pencal enables easy parallelization through the specification of the number of cores as argument within its functions. Reference: Signorelli, M. (2023). pencal: an R Package for the Dynamic Prediction of Survival with Many Longitudinal Predictors. Preprint: arXiv.2309.15600

Mirko Signorelli, Leiden University

I work as assistant professor of Statistics at Leiden University.
My research focuses on the development of new statistical models, the creation of R packages (neat, pencal, ptmixed, success), and the application of statistics to biomedical problems. I teach courses on R, computational statistics, and longitudinal data analysis.