A Comprehensive Overview of Autoencoder Algorithms to Leverage the Diagnosis of Complex Diseases

Abstract

Recent advances in high-throughput sequencing technologies such as whole genome sequencing, single-cell, and others, have boosted the understanding of complex diseases. However, extracting biological meaning using the data generated by these methods is not straightforward. Various analysis techniques, including machine learning algorithms, have been proposed recently. These techniques have recently proven to be beneficial in the medical field. Unsupervised learning methods using neural networks, such as autoencoders (AEs) or variational autoencoders (VAEs), have shown promising results among such approaches. Several applications have been presented on various types of data and in different contexts, spanning from cancer to healthy patient tissues. In this book chapter, we discuss how AEs and VAEs have been used in biomedical settings. Specifically, here we discuss their current applications and the improvements achieved in the diagnostic and survival of patients.