Visible neural networks for multi-omics integration: a critical review

Abstract

Biomarker discovery and drug response prediction is central to personalized medicine, driving demand for predictive models that also offer biological insights. Biologically informed neural networks (BINNs), also known as visible neural networks (VNNs), have recently emerged as a solution to this goal. BINNs or VNNs are neural networks whose inter-layer connections are constrained based on prior knowledge from gene on-tologies pathway databases. These sparse models enhance interpretability by embedding prior knowledge into their architecture, ideally reducing the space of learnable functions to those that are biologically meaningful. In this systematic review--the first of its kind-- we identify 86 recent papers implementing such models and highlight key trends in architectural design decisions, data sources and methods for evaluation. Growth in popularity of the approach is apparently mitigated by a lack of standardized terminology, tools and benchmarks.