From Digital Twin to Virtual Human Twin: A “Moonshot” Project in Digital Health

1. Academic Background and Research Motivation

Currently, the global healthcare system continues to face numerous unmet clinical and social needs, which manifest as a lack of treatment options, insufficient and expensive medical resources, lengthy waiting times, and inadequate attention to vulnerable groups such as children and those with rare diseases (unmet needs). Despite ongoing advancements in understanding the physiological mechanisms of health and disease, as well as the continuous emergence of new diagnostic and therapeutic techniques, the universality, efficiency, and personalization of healthcare services still fall short. As a result, the medical and industrial communities are actively exploring digital and informational approaches to drive transformation.

Similar to how the Human Genome Project decoded the entirety of human genetic information, the IUPS Physiome Project was the first to propose the idea of a “systematic digital dynamic model of human physiology”—constructing a virtual “digital human” model encompassing all known aspects of human pathophysiology. Subsequently, the EU-led “Virtual Physiological Human” (VPH) initiative further emphasized the potential of these computer models as Clinical Decision Support Systems (CDSS), promoting the initial commercial adoption of personalized medicine and “Digital Twin” (DT) technology.

The concept of the digital twin originated in the manufacturing sector, where computer models and real-time data create a virtual mirror of physical entities. In recent years, this idea has been applied across fields, leading to the development of “Digital Twin in Healthcare” (DTH)—patient-specific models that integrate diverse physiological data, medical imaging, and mechanistic knowledge to accurately predict disease progression and treatment responses. For example, the HeartFlow FFRct product leverages patient imaging data in a computational fluid dynamics model to non-invasively replace invasive measurements in coronary stenosis diagnostics, serving as a prototypical first-generation digital twin healthcare product.

Although some solutions have found their way into clinical practice, the “digital twin revolution” predicted for healthcare is still far from fully realized. The main reason lies in the fact that most current products focus on single, easily implemented components and lack systematic, scalable collaborative platforms. This results in significant obstacles such as data silos, limited model reuse, barriers to interdisciplinary collaboration, and underdeveloped market mechanisms. Thus, the authors propose the ambitious blueprint of the “Virtual Human Twin” (VHT) as a “moonshot” project in digital health, aiming to fundamentally drive innovation and consensus in the field.

2. Source and Authors of the Paper

This is a position paper co-authored by Marco Viceconti, Maarten de Vos, Sabato Mellone, and Liesbet Geris, who represent top European universities and medical institutions including the University of Bologna in Italy, Katholieke Universiteit Leuven, and the University of Liège in Belgium. The paper was published in the IEEE Journal of Biomedical and Health Informatics, Volume 28, Issue 1, January 2024, and is an open-access publication. It is supported by the European Commission’s H2020 project and the Horizon Europe EDITH coordination and support action, aiming to shape future research planning for digital health in Europe.

3. Main Thesis and Structure of the Paper

3.1 Key Terminology

• In Silico Medicine: Refers to all disease prevention, diagnosis, prognosis, and treatment modeling and simulation performed using computational methods. Models may be data-driven (AI predictions) or knowledge-driven (physics/mechanism-based predictions), each with pros and cons, and both reliant on hypotheses and evidence.
• Digital Twin in Healthcare (DTH): Patient-specific predictive models based on individual data, providing high-credibility decision support. Unlike traditional population-based models, DTH highlights personalization and real-time updating.
• Virtual Physiological Human (VPH): An EU-initiated technological framework aiming to build a collaborative model of human pathophysiology to serve both research and clinical applications.
• In Silico Trials: Clinical trials conducted using large numbers of digital twin models, with the goal of replacing or reducing the need for human or animal experiments in drug and medical device development.
• Intended Use and Context of Use: Respectively refer to the clinical decision support purpose of software-based medical devices, or the technical objective and practical application scenario in in silico trials.
• Knowledge-Driven and Data-Driven Models: The former are theory and mechanism-based, the latter rely primarily on large-scale statistics and AI algorithms. The two approaches generally need to be integrated.

3.2 Challenges—Why Digital Twins Struggle to Take Hold

The “In Silico World Community of Practice” (ISW_CoP) has systematically summarized the seven main barriers hindering the widespread adoption of digital twin technologies in healthcare:

1. Lack of Advanced Models: Especially in multi-scale, multi-system modeling, algorithms and knowledge struggle to reflect individual physiological variability accurately.
2. Lack of Available, Representative Data for Development and Independent Validation: Data sharing is severely constrained, strict data privacy laws apply, and high-quality, annotated datasets are scarce.
3. Unclear Regulatory Pathways: Computational simulation evidence is not fully accepted for the approval of new drugs and medical devices; existing standards (e.g., ASME VV-40:2018) need to be fully adapted for data-driven models, and regulatory cycles are long.
4. Insufficient Stakeholder Awareness: From patients to policymakers, there is a lack of unified and clear information and evaluation frameworks regarding the opportunities and risks of digital twins.
5. Poor Scalability and Efficiency: Current models require significant computational power and lack safe and convenient environments.
6. Shortage of Skilled Professionals: Training mechanisms for relevant technical talents are incomplete, leading to shortages in industry, research, and regulation.
7. Immature Business Models: Sustainable commercial mechanisms are lacking, with data/model sharing and monetization models needing exploration.

3.3 Virtual Human Twin (VHT): Blueprint for an Ecosystem

To address these challenges, the authors propose the VHT concept. VHT is not a singular, all-encompassing model, but a distributed collaborative infrastructure encompassing data, models, and standards, with the following key features:

1. Basic Orientation: VHT is not a super-model of everything, but an open ecosystem platform supporting the accumulation and interconnection of diverse data, models, and knowledge, greatly facilitating the development and validation of DTH (Digital Twin in Healthcare).

2. Data Objects and Model Objects: VHT’s data objects comply with FAIR (Findable, Accessible, Interoperable, Reusable) principles, prioritizing the inclusion of quantitative and individualized data, thoroughly annotating metadata. Model objects act as “data space crawlers,” enabling automatic configuration and computation of inputs and outputs.

3. Six-Dimensional Data Space:

   • Space: Mapping 0-3 dimensions onto a human anatomical template.
   • Time: Collection time, age, or lifespan range.
   • Credibility: Annotation of data validation and provenance, supporting stepwise certification.
   • Clustering: Enables flexible switching between individual and population data via multi-faceted clustering, e.g., by age, gender, health status.
   • Body Posture and Measurement Scales: Compatible with mechanical modeling and elastic registration for multi-scale alignment.
   • Other Dimensions: Fine definitions of data granularity and range, facilitating researchers’ search and utilization.

4. Model Orchestration and Remote Execution: Supports logical transfer between multiple models, automatic batch processing, and dedicated orchestration libraries for strongly coupled models (e.g., real-time interactive computation).

5. Ecosystem Role Assignment and Operation Mechanism: VHT serves clinical, research & development, and regulatory needs, and connects data/model providers to end users. Its structure supports flexible data storage/model migration, addressing legal, technical, and commercial constraints. The ecosystem may start with public funding and pre-competitive mechanisms, gradually move towards market operation, and ultimately form a self-sustaining, open collaborative platform.

3.4 Technical and Application Evaluation

Scientific Innovation and Application Prospects:

• Breaking Data Silos and Model Barriers: Based on high-quality sharing of data and models, VHT fosters interdisciplinary collaboration and accelerates the development of novel personalized healthcare tools.
• Accelerating In Silico Trial Adoption: Significantly reduces reliance on human and animal experiments, saving costs and shortening development cycles.
• Establishing Clear Regulatory and Validation Mechanisms: Open and independent collections for data and model validation will assist regulatory agencies and market participants in speeding up technology identification and standard setting.
• Information Platform for All Stakeholders: Patients, clinicians, researchers, industry, and regulators can all use VHT to access knowledge, test models, and share best practices.
• Empowering Integration of AI and Mechanistic Models: “Physics-driven machine learning” methods facilitate the acceleration of models and the analysis and prediction of complex pathophysiological mechanisms.
• Talent Development and Educational Innovation: Digital twin and virtual human templates offer vivid simulation and experimentation environments for medical education, potentially cultivating a new generation of interdisciplinary biomedical engineers.

Key Insights and Breakthroughs: - VHT is not a solution to specific diseases or single issues, but a foundational infrastructure supporting progress in multiple areas. - Emphasizes “model reusability, data interoperability, and standardization,” aiming to establish an open science foundation akin to biological databases. - Clearly proposes directions and standards for multi-scale, multi-system modeling, addressing the limitations of current single-organ, single-time-scale approaches.

Business and Social Impact: - VHT will promote the circulation and value addition of data, models, and computational resources, creating new business models and market chains for academia and industry. Data and model providers will receive fair incentives and rights protection, effectively driving prosperity in the digital health sector.

Ethical and Legal Challenges: - Although this paper does not elaborate on ethical and legal issues, it stresses the need for compliance with European and global regulations, which will be gradually refined in subsequent consensus processes.

4. Conclusion: Significance and Outlook

This paper, as the declaration and blueprint for the “Virtual Human Twin” project in digital health, marks a shift in medical digitalization from isolated breakthroughs to systematic and ecosystem-based development. It opens up new research horizons for academia while presenting new ideas for collaboration, innovation, and sharing among industry, regulators, and society at large. The ultimate goals of personalized and universal healthcare cannot be achieved through isolated institutions, traditional models, or fragmented data alone, but must rely on large-scale, open, collaborative platforms such as VHT to drive industrial upgrade and scientific revolution.

With the advancement of EU projects like EDITH and the construction of the VHT ecosystem, this “moonshot” initiative has the potential to drive digital transformation in healthcare over the next five to ten years, achieving a historic leap from scientific exploration to clinical implementation and from data silos to collaborative success.