# Organs-on-Chips in Drug Development: Engineering Foundations, Artificial Intelligence, and Clinical Translation

## TL;DR

This review explores how organ-on-a-chip technologies, combined with AI and advanced sensing, are improving drug development by modeling human physiology and enabling more accurate predictions of drug safety and efficacy.

## Contribution

The paper provides a comprehensive synthesis of recent advances in organ-on-a-chip technologies and their integration with AI for drug development and regulatory applications.

## Key findings

- Cardiac OoCs achieved AUROC ≥ 0.85 for torsadogenic risk classification.
- Renal chips improved prediction of transporter-mediated clearance compared to conventional assays.
- Platforms increasingly incorporate vascularization, immune components, and organoid hybrids with real-time biosensing.

## Abstract

Organ-on-a-chip (OoC) technologies, also termed microphysiological systems (MPSs), integrate microfluidics, engineered biomaterials, human-derived cells, and on-chip biosensing to model human physiology in microscale devices that deliver quantitative, time-resolved readouts. This review surveys the 2010–2025 literature, emphasizing how sensing, standardized sampling, and analytics enable clinical concordance and fit-for-purpose regulatory use. We synthesize advances in (i) materials, fabrication, and microfluidic design; (ii) organ- and disease-focused case studies; and (iii) translational benchmarks that align chip outputs with clinical pharmacokinetics, toxicology, and biomarker datasets. Across organ systems, platforms increasingly incorporate vascularization, immune components, and organoid hybrids, paired with real-time measurements of barrier integrity, metabolism, electrophysiology, and secreted biomarkers using impedance (TEER), electrochemical, and optical modalities. Representative benchmarking studies report cardiac OoCs achieving AUROC ≥ 0.85 for torsadogenic risk classification, and renal chips improving prediction of transporter-mediated clearance relative to conventional in vitro assays. We summarize validation approaches and regulatory developments relevant to new approach methodologies, including the FDA Modernization Act 2.0, and discuss how AI and multi-omics can automate signal and image analysis, harmonize cross-platform datasets, and support digital-twin workflows that couple OoC measurements to in silico models. Overall, biosensor-enabled OoCs are progressing toward quantitatively benchmarked platforms for safety pharmacology, ADME/PK–PD, and precision medicine.