DigitalTwinBio.comDigitalTwinBio.com

From flight hardware to living systems

The idea is an engineering one. Michael Grieves presented it in 2002 at the University of Michigan as a model for product lifecycle management, calling it a mirrored spaces model and later an information mirroring model; the name digital twin came from John Vickers at NASA, and Grieves and Vickers set out the mature three-part definition, a physical thing, its virtual counterpart, and the data flowing between them, in a 2017 chapter.1 Ten years later, a NASA and U.S. Air Force paper gave the version most often quoted in engineering: an integrated multiphysics, multiscale, probabilistic simulation of an as-built vehicle that uses the best available models, sensor updates and fleet history to mirror the life of its flying twin.2

Biology borrowed the word because the shape of the problem is the same. A body, a tumor, a bioreactor and a cell are all systems that are expensive to experiment on, partly observable, and different in every instance. What biology adds is that the counterpart is alive: it adapts while you are modeling it, and you rarely get to run the experiment twice.

The test: a two-way data link

The most useful definition is the plainest. A digital twin couples computational models with a physical counterpart to create a system that is dynamically updated through bidirectional data flows as conditions change.3 That word bidirectional does the work. A model that is built once and never fed measurements from its specific subject is a simulation, however good it is. A twin is tied to one heart, one patient, one production run, and is corrected by that subject's data as the data arrives, and in turn informs decisions about it.

The same National Academies study, with a companion workshop devoted to biomedical research,4 is also honest about what is missing: bioethics, the handling of measurement error and timing variation in clinical data, metadata good enough to let datasets be harmonized, and the safety of any closed loop in which a model influences care.3 Those are the real open problems, and they are worth naming plainly.

The scales biology twins at

There is no single biological digital twin. There is a stack of them, and the name covers the whole stack:

• The cell. A whole-cell model published in 2012 reproduced the life cycle of Mycoplasma genitalium from 28 integrated sub-models spanning replication, transcription, translation, metabolism and division, and was the first to represent every characterized gene function of an organism, 401 genes in all.5 A 2020 follow-on carried the approach to Escherichia coli and used the simulation to cross-check heterogeneous published datasets against each other.6
• The organ. Cardiology has the most developed position, combining mechanistic electrophysiology and mechanics with data-driven methods in what a European Society of Cardiology paper calls the digital twin for precision cardiology.7 The underlying engineering is ordinary finite-element work: a four-chamber heart built from CT and MR images, solved for coupled electrical and mechanical behavior.8 Vessel trees, lungs and kidneys follow the same pattern.
• The process. The most literal twin in the life sciences is industrial. Bioprocess twins of mammalian cell culture pair in-line sensors with mechanistic models, computational fluid dynamics for mixing and gas transfer, kinetics for growth and nutrient depletion, plus data-driven layers on top.9 Here the live feed genuinely exists, which is why manufacturing is ahead of the clinic.
• The patient. Multi-omics, wearables and electronic records feed statistical and learned models of one person. The promise is real and so is the list of barriers: heterogeneous data, interpretability, cost, equity, and the plain fact that clinical validation and health technology assessment come before adoption.23 Public programs are pushing on it, including an NCI and Department of Energy ideas lab on building a cancer patient digital twin.24

Mechanistic modeling already has a regulatory home

Long before anyone said twin, pharmacology was building mechanistic models of living systems. Quantitative and systems pharmacology was named as a discipline in a 2011 NIH white paper that argued pharmacology should move past drug and target interactions to a quantitative account of drug action across scales.10 Physiologically based pharmacokinetic modeling, which predicts where a compound goes in the body from physiology and compound properties, is mature enough to have its own reporting expectations: a final FDA guidance on the format and content of PBPK analyses,11 and an EMA guideline on reporting PBPK modeling and simulation adopted in December 2018.12

The umbrella term is model-informed drug development. It now has an international guideline, ICH M15, which defines MIDD as the strategic use of computational modeling and simulation that integrates nonclinical and clinical data, prior information and knowledge to produce evidence; it was endorsed as a draft in November 2024 and adopted at step 4 in February 2026.13 In the United States, sponsors can take a modeling question to the agency through the MIDD paired meeting program, a performance goal under PDUFA VII in which a small number of requests each quarter receive an initial and a follow-up meeting on the same issue.14

In silico, named in law

For decades United States drug law spoke of preclinical tests including tests on animals. The FDA Modernization Act 2.0, enacted as Section 3209 of the Consolidated Appropriations Act, 2023 and signed on 29 December 2022, replaced that phrase in Section 505 of the Federal Food, Drug, and Cosmetic Act with nonclinical tests, and defined those to include in vitro, in silico and in chemico methods alongside animal tests.15 Computer modeling is now named in the statute. Animal testing was not banned; the law permits rather than requires.

In April 2025 the FDA published a roadmap for reducing animal testing in preclinical safety studies, beginning with monoclonal antibodies and naming new approach methodologies that include in silico and computational models alongside organoids and organ-on-chip systems.16 The roadmap does not use the phrase digital twin, and it would be wrong to put that word in an agency's mouth. What it does show is the direction of travel for the evidence these models are expected to carry.

Credibility is the whole game

A model that influences a decision has to earn that influence. ASME V&V 40-2018, released in November 2018 and written with device manufacturers, simulation developers and the FDA in the room, sets the framework: the verification and validation evidence should be proportionate to how much the model is relied on and to the consequence of the decision being wrong.18 The FDA's final 2023 guidance on assessing the credibility of computational modeling and simulation in device submissions builds a nine-step process on top of that standard for physics-based and mechanistic models.17

Biology makes this harder than aerospace does. You cannot rerun a patient, validation data is sparse and noisy, and uncertainty has to be quantified rather than assumed away, which is why verification, validation and uncertainty quantification for medical twins is an active research subject of its own.19 The same critique explains why most twins are still one-off builds: scaling them from artisanal to industrial needs reusable model libraries, data pipelines and uncertainty machinery, not more bespoke solvers.20 The in silico trials community has been working the regulatory side of that problem for years, through published work on verification and validation for regulatory evaluation21 and through European projects set up to lower the barriers to adoption.22

Data is the binding constraint

A twin needs a continuing feed about one specific subject, which is exactly the data that is most regulated and most fragmented. In Europe the European Health Data Space regulation entered into force on 26 March 2025 and applies from 26 March 2027, with the chapter covering secondary use for research, policy and regulatory decisions, on anonymised or pseudonymised data, applying from March 2029.25 Consent, provenance, metadata quality and the right to withdraw are not side issues for a biological twin. They decide whether one can exist at all.

Who gathers around the name

The sources above are a fair map of who might want a category name like this:

• Simulation and modeling software companies in the life sciences, who need a short, literal product name.
• Pharma and biotech modeling groups doing QSP, PBPK and pharmacometrics under the MIDD umbrella.13
• Contract research organizations offering nonclinical in silico services, now a named category in United States law.15
• Medical device and diagnostics teams submitting computational evidence under the 2023 credibility guidance.17
• Academic spin-outs and consortia from the in silico medicine community.22
• Bioprocess and biomanufacturing software teams building process twins for cell culture.9
• Publishers, courses and conferences: a field this young needs a place to explain itself.

Digital twin, virtual patient, in silico model

The words are not interchangeable, and the difference is worth knowing before you build on the name. An in silico model is any computational representation of a biological system. A virtual patient is a model individual, often one of thousands in a synthetic cohort for an in silico trial, and need not correspond to a real person.21 A digital twin is narrower and stricter: it is tied to one real counterpart and kept in step with it by data.3 DigitalTwinBio.com sits above all three, which is the useful position for a platform, a research group or a publication: broad enough to hold the whole field, specific enough that nobody has to ask what it means.

1. Grieves M, Vickers J. Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems.In: Transdisciplinary Perspectives on Complex Systems, Springer, 2017
2. Glaessgen EH, Stargel DS. The Digital Twin Paradigm for Future NASA and U.S. Air Force Vehicles.NASA Technical Reports Server 20120008178, AIAA 2012-1818, 2012
3. Foundational Research Gaps and Future Directions for Digital Twins.National Academies of Sciences, Engineering, and Medicine, 2024
4. Opportunities and Challenges for Digital Twins in Biomedical Research: Proceedings of a Workshop.National Academies, workshop held 30 January 2023
5. Karr JR et al. A Whole-Cell Computational Model Predicts Phenotype from Genotype.Cell 150(2):389-401, 2012
6. Macklin DN et al. Simultaneous cross-evaluation of heterogeneous E. coli datasets via mechanistic simulation.Science 369(6502):eaav3751, 2020
7. Corral-Acero J et al. The ‘Digital Twin’ to enable the vision of precision cardiology.European Heart Journal 41(48):4556-4564, 2020
8. Baillargeon B, Rebelo N, Fox DD, Taylor RL, Kuhl E. The Living Heart Project: a robust and integrative simulator for human heart function.European Journal of Mechanics A/Solids 48:38-47, 2014
9. Park S-Y et al. Bioprocess digital twins of mammalian cell culture for advanced biomanufacturing.Current Opinion in Chemical Engineering 33:100702, 2021
10. Emergence of Quantitative and Systems Pharmacology: A White Paper.NIH National Institute of General Medical Sciences, October 2011
11. Physiologically Based Pharmacokinetic Analyses — Format and Content. Guidance for Industry.U.S. Food and Drug Administration, final guidance, 2018
12. Guideline on the reporting of physiologically based pharmacokinetic (PBPK) modelling and simulation (EMA/CHMP/458101/2016).European Medicines Agency, adopted 13 December 2018
13. Harmonised ICH M15 Guideline: General Principles for Model-Informed Drug Development, adopted.International Council for Harmonisation, step 4, February 2026
14. Model-Informed Drug Development Paired Meeting Program.U.S. Food and Drug Administration, PDUFA VII
15. Consolidated Appropriations Act, 2023, Public Law 117-328, Section 3209 (Animal testing alternatives): the FDA Modernization Act 2.0.Approved 29 December 2022; 136 Stat. 5821-5822
16. Roadmap to Reducing Animal Testing in Preclinical Safety Studies.U.S. Food and Drug Administration, April 2025
17. Assessing the Credibility of Computational Modeling and Simulation in Medical Device Submissions.U.S. Food and Drug Administration, final guidance, November 2023
18. ASME V&V 40-2018: Assessing Credibility of Computational Modeling through Verification and Validation: Application to Medical Devices.American Society of Mechanical Engineers, 2018
19. Sel K et al. Survey and perspective on verification, validation, and uncertainty quantification of digital twins for precision medicine.npj Digital Medicine 8, 2025
20. Niederer SA, Sacks MS, Girolami M, Willcox K. Scaling digital twins from the artisanal to the industrial.Nature Computational Science 1:313-320, 2021
21. Viceconti M et al. In silico trials: verification, validation and uncertainty quantification of predictive models used in the regulatory evaluation of biomedical products.Methods 185:120-127, 2021 (PMC7883933)
22. In Silico World: Lowering barriers to ubiquitous adoption of In Silico Trials.European Commission CORDIS, Horizon 2020 grant agreement 101016503
23. De Domenico M et al. Challenges and opportunities for digital twins in precision medicine from a complex systems perspective.npj Digital Medicine 8, 2025
24. NCI-DOE Collaboration 2020 Ideas Lab: Toward Building a Cancer Patient “Digital Twin”.NIH Office of Data Science Strategy, 2020
25. Regulation (EU) 2025/327 establishing the European Health Data Space.EUR-Lex, in force 26 March 2025