Integration of microphysiological systems with computational and digital twin modeling for pharmaceutical development: a systematic review.

Abstract

Background: The pharmaceutical industry faces a productivity crisis, with only ∼10% of candidates reaching approval and development costs exceeding $1-2 billion per drug. Traditional preclinical models, particularly animal studies, show limited predictive validity for human outcomes. Microphysiological systems (MPS), including organ-on-chip devices, integrated with digital twin computational modeling, offer a promising route to improve translational prediction, yet no systematic assessment exists.

Objectives: To identify, evaluate, and synthesize the published literature on integrating MPS with digital twin computational modeling for pharmaceutical development.

Methods: Following PRISMA 2020 guidelines (PROSPERO: CRD420251274941), PubMed and Europe PMC were searched from January 2010 to December 2025. Studies were included if they integrated MPS platforms with computational models (PBPK, QSP, IVIVE, PK/PD, or machine learning). Two-stage automated screening with independent verification by two researchers was applied. Data extraction captured organ systems, model types, platforms, validation approaches, pharmacokinetic parameters, software, and economic considerations.

Results: Of 2,044 records, 123 studies met inclusion criteria. Publication activity grew exponentially (CAGR ∼35%), with 38.2% published in 2024-2025. Liver was the most represented organ (30.9%), followed by vasculature (19.5%), gut (18.7%), and immune components (17.1%). PBPK was the predominant computational approach (18.7%), followed by PK (12.2%), IVIVE (12.2%), and machine learning (9.8%). Commercially, Emulate led (55.8%), then Mimetas (23.3%) and CN Bio (19.2%); custom/academic platforms comprised 74.2%. Validation was robust, with 95.8% discussing clinical data comparisons and 77.5% reporting average fold error. Only liver-PBPK and liver-PK qualified as established combinations (>5 studies), with many organ-model pairings unexplored. Multi-organ systems showed accelerating adoption after 2021. Economic themes were discussed qualitatively, but formal health economic analyses were absent.

Conclusion: MPS-digital twin integration has matured from an emerging concept into a research paradigm with demonstrated translational value. Validation practices are strong and aligned with regulatory expectations, supported by the FDA Modernization Act 2.0 and the ISTAND program's acceptance of Liver-Chip technology. Liver-focused applications are the most mature and should serve as a template for expansion. Priority gaps include kidney, lung, and brain MPS-computational integration, multi-organ systems, and rigorous health economic analyses. Continued standardization and regulatory engagement will be essential to realize this approach's potential.