Fellowship project title: Harnessing longitudinal data to improve the prediction of survival outcomes: making the R package pencal more efficient, accessible and visible
The goal: Improve the efficiency, usability and visibility of the R package pencal. pencal makes it possible to improve the accuracy of predictions of survival probabilities through the use of information from several variables measured longitudinally over time.
Accurately estimating survival probabilities is important to adequately inform patients and their families, and to guide treatment decisions. Nowadays, more and more data are measured repeatedly over time; these longitudinal data can contain valuable information about the patient’s disease progression.
Currently, the estimation of survival probabilities is usually based on data gathered at either the first or the last visit available from the patient, ignoring information from all other visits. Although in principle information from multiple visits could be used to improve prediction accuracy substantially, this is rarely done in practice due to the lack of software tools that can handle many longitudinal covariates as predictors of survival.
This project aims to improve the efficiency, usability and visibility of the R package pencal, which implements a statistical method called Penalized Regression Calibration (PRC, Signorelli et al., 2021). PRC makes it possible to use many longitudinally-measured covariates as predictors of survival, thereby leveraging the full potential of longitudinal data and improving prediction accuracy.
Reference: Signorelli, M., Spitali, P., Al-Khalili Szigyarto, C., The Mark-MD Consortium, Tsonaka, R. (2021). Penalized regression calibration: A method for the prediction of survival outcomes using complex longitudinal and high-dimensional data. Statistics in Medicine, 40 (27), 6178-6196.