The reality revealed by 598 articles in Web of Science makes visible the gap between what digital twin technology promises for healthcare and what it actually produces in practice.
The value of healthcare, in Porter’s (2010) classical definition, is the ratio of patient outcomes achieved to the resources spent to achieve those outcomes. This simple fraction actually contains the most difficult equation of modern healthcare management: increasing the numerator, namely clinical outcomes, while reducing the denominator, namely resource use. For decades, health systems have positioned these two ends as opponents. Improvement programs have generally focused either on patient outcomes or on operational efficiency, while holistic approaches that treat both as two sides of the same decision have remained rare.
Digital twin technology has entered the literature over the last five years as one of the rare tools capable of radically changing this equation. This technology, which can create a computational replica of a physical entity updated with real-time data, promises both physiological personalization at the patient level and operational optimization at the hospital level (Laubenbacher et al., 2022; Vallée, 2024). Theoretically, the picture is attractive: while a patient-level cardiac digital twin can personalize treatment decisions and reduce mortality (Gillette et al., 2021), the operational digital twin of the same hospital can optimize bed capacity and staff allocation, thereby reducing costs (Karakra et al., 2025). If these two layers are connected, a rare output that we may call “dual value” in the healthcare management literature emerges.
However, the systematic analysis of 598 peer-reviewed digital twin articles published between 2019 and 2026 in the Web of Science Core Collection shows that this promise has not yet been fulfilled. This text aims to discuss the anatomy of this paradox, its causes, and the conditions under which it can be overcome.
What the Numbers Say: The Missing Link of One and a Half Percent
The distribution of value orientation among the 598 articles shows that the digital twin literature in healthcare is still in a technical-conceptual stage. Of the articles, 77.3% do not define any value criterion; they merely provide methodological demonstrations, conceptual frameworks, or technical feasibility reports. Among the 136 articles that define a value criterion, 75 focus only on patient clinical outcomes, such as mortality, survival, complications, and quality of life, while 52 focus only on operational efficiency, such as cost, resources, and waiting time. The number of articles that simultaneously optimize patient outcomes and operational efficiency within the same model is only nine. Their proportion within the total is 1.5%, and their proportion among value-oriented studies is 6.6%. This figure is quantitative evidence of the “dichotomy” identified by Xames et al. (2025) between engineering and healthcare communities: clinical digital twins are fed by the culture of biomedical engineering, whereas operational digital twins are fed by the culture of industrial engineering; the two streams do not converge in practice.
What is even more striking is that dual value has already been shown to be technically possible. Hiraoka et al. (2025) reported that, by applying production efficiency principles to operating room processes in robotic knee prosthesis surgery, they simultaneously improved both patient outcomes and operating room throughput. Silva-Aravena et al. (2025), in a reinforcement learning-based digital twin developed for MRI scheduling, produced Pareto frontier solutions that jointly optimized clinical urgency scores and machine utilization rates. Fuchs et al. (2023) showed how the value-based healthcare framework could be integrated with a digital twin by combining PROMs, patient-reported outcome measures, CROMs, clinician-reported outcome measures, and multidisciplinary team cost analysis into a single benchmarking model for sarcoma patients. In other words, the problem is not that “it cannot be done”; the problem is that “it is not being done.”
The Three Structural Causes of the Paradox
The rarity of dual value can be interpreted along three axes. The first is the disciplinary silo effect. Clinical digital twin studies are mostly published in biomedical engineering journals such as Medical Image Analysis, Computer Methods and Programs in Biomedicine, and Frontiers in Physiology, while operational digital twin studies appear in healthcare service systems and operations research outlets. These two communities do not have common conferences, common terminology, or a common valuation framework. This disconnection has not been resolved in most studies, even though more permeable journals such as NPJ Digital Medicine have emerged.
The second cause is methodological. Dual value optimization is, by its nature, a multi-objective problem; it requires multi-objective optimization methods that search for the Pareto frontier or multi-reward reinforcement learning architectures. Yet, in the dataset of 598 articles, multi-objective optimization was used in only seven studies, corresponding to 1.2%, and reinforcement learning in only ten studies, corresponding to 1.7%. By contrast, simulation-based approaches such as DES, ABS, and Monte Carlo methods appear in 237 articles, corresponding to 39.6%, while physics and machine learning hybrid models appear in 176 articles, corresponding to 29.4%, forming the mainstream of the literature. The current methodological distribution indicates that the tools capable of asking the dual value question are not being used sufficiently.
The third cause is the difficulty of data integration. The fragmented structure of healthcare information systems makes it seriously difficult to feed clinical data, such as EHR, PACS, and laboratory results, and operational data, such as ERP, hospital information systems, and equipment IoT streams, into a single model engine. As emphasized by Ali et al. (2023) in their review of the federated learning literature, health data sharing architectures have not yet reached the maturity required to routinely support the real-time integration demanded by operational digital twin architectures.
The Operating Room: The Natural Test Bed of Dual Value
The unit-level analysis of the dataset shows that the operating room is the most suitable test bed for dual value. While 8.3% of operating room studies produce dual value, this rate decreases to 5.1% in outpatient clinics, 4.2% in radiology, and 2.6% in intensive care. No dual value-producing study was found in emergency departments, pharmacies, or rehabilitation units. There is a structural reason why the operating room stands out: the surgical process combines high-cost resources, such as surgeon time, anesthesia teams, and robotic platforms, with direct clinical outcome indicators, such as complications, functional outcomes, and survival, within a single decision matrix. This dual dependency provides a natural mathematical basis for dual value modeling. The study by Hiraoka et al. (2025) concretely demonstrates how this basis can be used productively.
Intensive care is the second priority candidate. Appuhamilage et al. (2025) optimized critical care workflows with a discrete event simulation-based digital twin framework, yet no model in the literature has empirically tested the mortality-resource trade-off. Considering that intensive care has ideal conditions for dual value testing, namely high cost, short time horizon, and measurable clinical outcomes, this gap is expected to be filled rapidly within the next three to five years.
A Four-Layer Architectural Framework
To overcome the paradox, a four-layer framework can be proposed based on the shared architectural features of successful dual value examples in the literature. The first layer is the data fusion layer, where clinical data sources, such as EHR, PACS, and laboratory systems, are integrated with operational data sources, such as ERP, hospital information systems, and equipment IoT, through open standards such as FHIR R4. The cyber-physical system architecture of Shaikh et al. (2023) provides the conceptual basis for this layer. The second layer is the dual model engine, where the clinical digital twin (Camps et al., 2024; Gillette et al., 2021) and the operational digital twin (Karakra et al., 2025) generate parallel outputs using the same data stream. The third layer is the dual value optimizer, in which clinical and operational outputs are optimized simultaneously. The reinforcement learning approach of Silva-Aravena et al. (2025) offers a concrete prototype of this layer, while the closed-loop control logic in ReplayBG developed by Cappon et al. (2023) provides the mechanism for continuous updating. The fourth layer is the explainable decision interface, which visualizes scenarios along the Pareto frontier. The principle of ethical transparency emphasized by Huang et al. (2022) is decisive in the design of this interface.
Organizational Design: Three Principles That Make the Architecture Work
No matter how elegant the architectural framework is, it cannot work in practice without an organizational foundation. Three design principles establish this foundation. The first principle is the creation of a cross-functional value team. Biomedical engineers, operations researchers, clinicians, and health economists must be brought together within the same project management team. The analysis conducted by Xames et al. (2025) using the CFIR 2.0 framework clearly shows that the disconnection between the two communities is a fundamental cause of implementation failure. The second principle is the institutionalization of a dual dashboard. Clinical outcome indicators, such as complication rates, survival, PROMs, and CROMs, and operational indicators, such as bed utilization, throughput, and cost, should be presented side by side on a single management dashboard. The system should not allow one indicator to improve at the expense of the other. The sarcoma benchmarking study by Fuchs et al. (2023) is the most mature example of this approach in the literature. The third principle is embedding the VBHC framework, particularly Porter’s IPU, integrated practice unit model, as an anchor point at the design stage of digital twin projects. The study by Fuchs et al. (2023) remains the only example in the literature that directly integrates VBHC with a digital twin; the spread of this principle represents the natural direction of the field’s evolution.
Mature Fields, Niche Fields, and the Ethical Dimension
The taxonomy of 21 main fields and 36 subfields in the 598-article dataset maps the real-life problems that digital twin technology addresses in healthcare. Healthcare service systems (n = 193), intensive care (n = 150), geriatrics (n = 148), and the cardiovascular system (n = 148) are the most mature research clusters. This maturity can be explained by both clinical urgency and data availability. All four fields produce intensive data and involve high decision pressure. By contrast, niche areas such as dentistry (n = 8), chronic pain management (n = 8), obstetrics (n = 9), and 3D bioprinting (n = 11) have low frequency but high translational potential. In particular, digital twin-based decision support platforms developed for pediatric obesity, such as PODiaCarD, transdermal fentanyl release modeling in chronic pain management, and fetal heart digital twins stand out as pioneering research areas for the coming period.
The ethical and regulatory dimension must not be ignored. In the dataset, 58 articles directly address medical ethics, data privacy, and regulatory frameworks. Issues such as patient data protection, algorithmic bias, informed consent, and GDPR compliance emphasize that the integration of digital twin technology into clinical practice is not only a technical process but also a normative one. In the field of patient safety, applications involving the BowTie model, adverse event prediction, and safety barrier simulation are developing. Adding “ethical value” as the third dimension of the dual value paradigm is one of the most critical openings for the future research agenda.
Conclusion: The Paradox Can Be Overcome, and the Direction Is Clear
This analysis shows that digital twin technology is growing rapidly in healthcare, from 7 articles in 2019 to 189 articles in 2025, with 50.5% of all publications concentrated in 2025 and 2026. However, the field still experiences a paradox in which clinical and operational streams do not converge in practice. The dual value rate of one and a half percent is a concrete indicator of the gap between the promise and practice of the technology. At the same time, it is also clearly shown that the paradox can be overcome: the studies by Hiraoka et al. (2025), Silva-Aravena et al. (2025), Fuchs et al. (2023), and Wang et al. (2026) prove that dual value is technically possible.
The practical message for healthcare managers is clear. When digital twin investments are structured with the expectation of either clinical gain alone or operational gain alone, the true transformative power of the technology does not emerge. Projects aiming to generate dual value should include a cross-functional team from the design stage, place the dual dashboard at the center of decision-making processes, and preserve the VBHC framework as the project’s anchor point. Operating rooms and intensive care units are the most suitable starting points for pilot implementations. When technical developments mature simultaneously with ethical and regulatory frameworks, when clinical validation studies increase, and when integration models are developed at the level of healthcare service systems, the real-life impact of the digital twin in healthcare will become visible.
The value equation of healthcare does not necessarily involve a tension between its numerator and denominator. This is not a structural feature of the system, but a historical choice. When digital twin technology is built with the right architecture and organizational foundation, it is one of the rare tools capable of reversing this choice.
References
Ali, M., Naeem, F., Tariq, M., & Kaddoum, G. (2023). Federated learning for privacy preservation in smart healthcare systems: A comprehensive survey. IEEE Journal of Biomedical and Health Informatics, 27(2), 778–789. https://doi.org/10.1109/JBHI.2022.3181823
Appuhamilage, G. D. K. K., Hussain, M., Zaman, M., & Khan, W. A. (2025). A health digital twin framework for discrete event simulation based optimised critical care workflows. NPJ Digital Medicine, 8(1), Article 376. https://doi.org/10.1038/s41746-025-01738-4
Camps, J., Berg, L. A., Wang, Z. J., Sebastian, R., Riebel, L. L., Doste, R., … Rodriguez, B. (2024). Digital twinning of the human ventricular activation sequence to clinical 12-lead ECGs and magnetic resonance imaging using realistic Purkinje networks for in silico clinical trials. Medical Image Analysis, 94, Article 103108. https://doi.org/10.1016/j.media.2024.103108
Cappon, G., Vettoretti, M., Sparacino, G., Favero, S. D., & Facchinetti, A. (2023). ReplayBG: A digital twin-based methodology to identify a personalized model from type 1 diabetes data and simulate glucose concentrations to assess alternative therapies. IEEE Transactions on Biomedical Engineering, 70(11), 3227–3238. https://doi.org/10.1109/TBME.2023.3286856
Fuchs, B., Schelling, G., Elyes, M., Studer, G., Bode-Lesniewska, B., Scaglioni, M. F., Giovanoli, P., Heesen, P., & Wong, D. V. (2023). Unlocking the power of benchmarking: Real-world-time data analysis for enhanced sarcoma patient outcomes. Cancers, 15(17), Article 4395. https://doi.org/10.3390/cancers15174395
Gillette, K., Gsell, M. A. F., Prassl, A. J., Karabelas, E., Reiter, U., Reiter, G., … Plank, G. (2021). A framework for the generation of digital twins of cardiac electrophysiology from clinical 12-leads ECGs. Medical Image Analysis, 71, Article 102080. https://doi.org/10.1016/j.media.2021.102080
Hiraoka, A., Swinnen, B., Vandeputte, A., Franssen, W., & Leirs, G. (2025). Optimizing operating room efficiency in robotic-assisted total knee arthroplasty through manufacturing efficiency principles. Journal of Experimental Orthopaedics, 12(2), Article e70283. https://doi.org/10.1002/jeo2.70283
Huang, P. H., Kim, K. H., & Schermer, M. (2022). Ethical issues of digital twins for personalized health care service: Preliminary mapping study. Journal of Medical Internet Research, 24(1), Article e33081. https://doi.org/10.2196/33081
Karakra, A., Fontanili, F., Lamine, E., & Lamothe, J. (2025). A novel application of discrete event simulation as a digital twin for monitoring patient pathways in a hospital in real time. Network Modeling Analysis in Health Informatics and Bioinformatics, 15(1), Article 15. https://doi.org/10.1007/s13721-025-00691-2
Laubenbacher, R., Niarakis, A., Helikar, T., An, G., Shapiro, B., Malik-Sheriff, R. S., … Glazier, J. A. (2022). Building digital twins of the human immune system: Toward a roadmap. NPJ Digital Medicine, 5(1), Article 64. https://doi.org/10.1038/s41746-022-00610-z
Porter, M. E. (2010). What is value in health care? New England Journal of Medicine, 363(26), 2477–2481. https://doi.org/10.1056/NEJMp1011024
Shaikh, T. A., Rasool, T., & Verma, P. (2023). Machine intelligence and medical cyber-physical system architectures for smart healthcare. Artificial Intelligence in Medicine, 146, Article 102692. https://doi.org/10.1016/j.artmed.2023.102692
Silva-Aravena, F., Morales, J., Jayabalan, M., & Sáez, P. (2025). Optimizing MRI scheduling in high-complexity hospitals: A digital twin and reinforcement learning approach. Bioengineering, 12(6), Article 626. https://doi.org/10.3390/bioengineering12060626
Vallée, A. (2024). Envisioning the future of personalized medicine: Role and realities of digital twins. Journal of Medical Internet Research, 26, Article e50204. https://doi.org/10.2196/50204
Wang, H. L., Chai, X. H., & Zhao, N. (2026). Digital twin virtual hospitals and rural health disparities: A six-country comparative study (2018–2024). Frontiers in Public Health, 14, Article 1741438. https://doi.org/10.3389/fpubh.2026.1741438
Xames, M. D., Topcu, T. G., Parker, S. H., Zagarese, V., & Epling, J. W. (2025). Evaluating CFIR 2.0 in identifying digital twin implementation challenges in healthcare: Bridging the dichotomy between engineering and healthcare communities. Frontiers in Digital Health, 7, Article 1611225. https://doi.org/10.3389/fdgth.2025.161122