Youden index-based methods to select and combine biomarkers and covariates.

C. J. Salaroli, M. C. Pardo

In -omics technologies, i.e. genomics (genes), transcriptomics (mRNA), and so on, often provide high-dimensional data, with thousands of features but only dozens of observations.
We propose PYE, a Penalized Youden index method to identify and combine the relevant biomarkers in disease classification, suitable in high-dimensional scenarios. Applications on synthetic and real datasets showed how PYE is capable to reach different preferable solutions, based on the considered penalization terms.
Once identified the biomarkers, we have been investigating the incremental contribution of covariates like physical attributes, lifestyle, alternative -omics data - also high-dimensional -.
We propose covYI, a method to select and combine covariates, identifying a covariate-specific cut-off point for every patient. The final outcomes are still under investigation, but the first results show that covYI is capable to lead to a significant increase in the classification performance.

Keywords: disease classification, high-dimensional data, biomarkers, covariates, selection and combination, penalized Youden index

Scheduled

GT04.BIO1 Invited Session. Complex Data
November 7, 2023  3:30 PM
CC3: Room 1

A penalized reduced rank proportional hazards model for age-related multi-morbidity event data

M. Rodríguez-Girondo

Graphical model inference with external network data

J. Jewson