Algorithms for and Challenges in the Analysis of Markers in Personalized Health Care

Abstract

Nowadays, the various omics disciplines such as genomics, proteomics, metabolomics, metagenomics, and transcriptomics generate a plethora of data. At the same time, a multitude of omics markers may be accompanied by a multitude of diseases. Hence, finding relationships between omics markers and disease in their early stages is a challenge that is at the very core of predictive or personalized medicine. In this chapter, an overview of algorithms for solving these problems of supervised learning is given, and challenges in this problem domain are discussed. Questions of learnability should be considered, and the quality and precision of the predictions should be assessed critically and quantitatively. Therefore, quality metrics for the assessment of the predictions are discussed as well.