Digital twins in healthcare

Current State

Digital twin (DT) technology, initially prominent in manufacturing, is increasingly applied in healthcare to create virtual replicas of patients, organs, or biological systems (Fuller et al., 2020; Sun et al., 2023). These "health digital twins" (HDTs) leverage real-time data from various sources, advanced analytics, and virtual simulations to enhance patient care (Vallée et al., 2023; Venkatesh et al., 2022). Key applications include personalized treatment planning, predictive analytics for disease, drug development, and surgical simulation (Acero et al., 2020; Björnsson et al., 2019). The technology is seen as a cornerstone for precision medicine, enabling therapies tailored to individual patient characteristics and physiological data (Acero et al., 2020; Thangaraj et al., 2024).

Strongest Evidence

Research highlights the potential of DTs to revolutionize healthcare by integrating multimodal patient data into mechanistic and statistical models (Thangaraj et al., 2024). For instance, in cardiology, DTs enable personalized simulations, disease risk prediction, and clinical trial augmentation (Acero et al., 2020; Thangaraj et al., 2024). Cloud-based frameworks are being developed to support elderly healthcare services, optimizing medical pathway planning and resource allocation (Liu et al., 2019). The concept is also being explored for drug discovery, where high-resolution patient models are computationally treated with thousands of drugs to find optimal therapies (Björnsson et al., 2019). Reviews consistently emphasize DTs' role in personalized diagnosis and treatment (Sun et al., 2023; Armeni et al., 2022).

Unresolved Uncertainties

Despite significant progress, several challenges impede the widespread adoption of DTs in healthcare. These include technical hurdles related to data integration from diverse sources, the complexity of modeling biological systems accurately, and robust validation methods for virtual replicas (Sun et al., 2023; Armeni et al., 2022). Regulatory frameworks and ethical considerations, such as data privacy, security, and algorithmic bias, also present substantial roadblocks (Popa et al., 2021; Armeni et al., 2022; Venkatesh et al., 2022). The seamless integration of AI, IoT, and big data, which are foundational to DTs, also brings its own set of socio-ethical issues (Popa et al., 2021; Barricelli et al., 2019).

Why the Topic Matters

Digital twins in healthcare hold immense promise for transforming medicine from a reactive to a proactive and personalized approach. By enabling precise simulations and predictions, DTs can lead to more effective treatments, reduced healthcare costs, and improved patient outcomes (Vallée et al., 2023; Armeni et al., 2022). The technology is crucial for realizing the vision of precision medicine, offering unprecedented opportunities for understanding disease progression, optimizing interventions, and advancing drug development (Acero et al., 2020; Björnsson et al., 2019). Addressing current challenges will unlock the full potential of connected care and redefine how health and chronic diseases are managed (Armeni et al., 2022).

Points of Agreement

• Digital twin (DT) technology is a transformative concept in healthcare, enabling personalized medicine through virtual replicas of patients, organs, or biological systems (Acero et al., 2020; Sun et al., 2023; Vallée et al., 2023).
• DTs integrate real-time data, advanced analytics, and virtual simulations to enhance patient care, facilitate predictive analytics, and optimize clinical operations (Vallée et al., 2023).
• The core components of a DT include physical entities, virtual models, data integration, and connections between the physical and virtual realms (Fuller et al., 2020).
• DTs are crucial for achieving precision medicine by providing tailored diagnoses and treatments based on extensive individual patient data (Acero et al., 2020; Sun et al., 2023).

Active Disagreements

• There is ongoing discussion regarding the standardization of DT architectures and interoperability across different healthcare systems and data sources.
• The optimal balance between data privacy, security, and the extensive data sharing required for comprehensive DT development is a subject of debate.
• Different approaches to validating the accuracy and reliability of DT models in diverse clinical scenarios are still being explored and compared.

Uncertain Claims

• The long-term clinical efficacy and cost-effectiveness of widespread DT implementation in routine healthcare remain to be definitively proven.
• The scalability of current DT frameworks to accommodate the vast and heterogeneous data from entire patient populations is uncertain.
• The precise ethical and regulatory frameworks required for the responsible deployment and governance of DTs in healthcare are still under development.

Evidence Gaps

• A lack of large-scale, prospective clinical trials demonstrating the tangible benefits and improved patient outcomes attributable to DT technology.
• Limited real-world evidence on the economic impact and return on investment for healthcare providers adopting DT solutions.
• Insufficient standardized methodologies for validating and verifying DT models across different medical specialties and patient cohorts.
• More research is needed on the human factors and user acceptance of DTs by both clinicians and patients.

The application of digital twin (DT) technology in healthcare is a rapidly evolving field, with strong evidence supporting its potential to revolutionize patient care and personalized medicine. DTs are defined as the seamless integration of data between a physical and virtual entity, enabling dynamic, data-driven virtual replicas (Fuller et al., 2020). In healthcare, this translates to creating virtual models of biological systems, organs, or individual patients, leveraging real-time data, advanced analytics, and virtual simulations (Vallée et al., 2023).

Claim 1: Digital twins enhance personalized treatment and predictive analytics in healthcare.

• Supporting Evidence: Multiple reviews highlight DTs' ability to offer personalized treatment plans based on individual characteristics and enable predictive analytics by integrating diverse patient data (Vallée et al., 2023; Sun et al., 2023). This aligns with the vision of precision medicine, where therapies are tailored to each patient using extensive individual data and powerful computational models (Acero et al., 2020). The technology facilitates customized diagnosis and treatment, moving beyond traditional approaches (Sun et al., 2023).
• Limitations: While the conceptual framework is strong, practical implementation faces challenges. These include the complexity of integrating vast amounts of heterogeneous data from various sources, ensuring the accuracy and real-time synchronization of the virtual twin with its physical counterpart, and validating the predictive power of these models in diverse clinical scenarios (Fuller et al., 2020; Barricelli et al., 2019).
• Evidence Strength: Strong. The consensus across multiple review papers and position papers indicates a clear understanding of the potential and a consistent vision for DTs in personalized healthcare.

Claim 2: Digital twin technology facilitates optimization of clinical operations and training.

• Supporting Evidence: DTs are recognized for their role in optimizing clinical operations and facilitating training and simulation within healthcare systems (Vallée et al., 2023). This extends beyond direct patient care to broader system-level improvements.
• Limitations: The evidence primarily discusses the potential and envisioned applications rather than presenting extensive empirical data on widespread successful implementations or quantifiable improvements in operational efficiency or training outcomes. Further research is needed to demonstrate the tangible benefits and return on investment in these areas.
• Evidence Strength: Moderate. While the concept is plausible and frequently mentioned, concrete evidence of widespread, validated impact is less robust compared to personalized treatment applications.

Digital twin (DT) technology is increasingly recognized as a transformative force in healthcare, moving beyond its origins in manufacturing (Fuller et al., 2020). Recent trends highlight its application in achieving precision medicine by enabling customized diagnoses and treatments for individual patients (Sun et al., 2023). This involves creating dynamic, data-driven virtual replicas of biological systems, organs, or entire patients (Acero et al., 2020).

Emerging methods focus on integrating real-time patient data from various sources with advanced analytics and virtual simulations to enhance patient care, facilitate predictive analytics, and optimize clinical operations (Vallée et al., 2023). For instance, in cardiology, DTs are envisioned to tailor therapies by capturing extensive patient data and leveraging powerful algorithms for learning and reasoning (Acero et al., 2020).

Despite significant progress, challenges persist, particularly in the seamless integration of diverse data, the robust validation of complex models, and addressing the ethical implications of using such detailed patient replicas (Sun et al., 2023). These areas represent ongoing research and development priorities to fully realize the potential of digital twins in healthcare.

Synthesis

Digital twins (DTs) in healthcare are virtual, dynamic models of a physical entity, such as a patient, organ, or physiological process, that are updated with real-time data (Fuller et al., 2020; Barricelli et al., 2019). By integrating data from sources like IoT sensors, medical records, and population studies, DTs create a high-fidelity replica that enables simulation, prediction, and optimization of health outcomes (Sun et al., 2023). The core vision is to shift from reactive to proactive and personalized medicine, where treatments and interventions can be tested in silico before being applied to the physical patient (Acero et al., 2020; Björnsson et al., 2019).

Methods and Materials

The creation of a health digital twin (HDT) relies on several key enabling technologies. The Internet of Things (IoT) provides the infrastructure for continuous, real-time data collection from wearables and other sensors. Cloud computing offers the necessary storage and processing power for the vast datasets involved (Liu et al., 2019). At the core are advanced computational models, which can be mechanistic (based on physiological principles) or statistical (driven by machine learning and AI) (Thangaraj et al., 2024). These models integrate multimodal data to simulate the behavior of the physical twin, often leveraging multi-agent systems to represent complex interactions (Croatti et al., 2020).

Applications

DTs are being explored across numerous healthcare domains. A prominent application is precision cardiology, where patient-specific heart models help diagnose, predict disease progression, and plan interventions (Acero et al., 2020; Thangaraj et al., 2024). They are also used to develop personalized treatment plans, optimize drug development, and simulate surgical procedures (Sun et al., 2023). Other applications include optimizing clinical operations and providing tailored healthcare services for the elderly (Liu et al., 2019; Vallée et al., 2023).

Limitations and Challenges

Despite their potential, the widespread implementation of DTs faces significant hurdles. Technical challenges include ensuring data quality, interoperability between different data sources, and the immense computational power required for high-fidelity simulations (Venkatesh et al., 2022). Furthermore, there are substantial regulatory, ethical, and social concerns regarding data privacy, security, and algorithmic bias (Armeni et al., 2022; Popa et al., 2021). Establishing a consensus on validation methods and clear regulatory pathways is critical for translating DT research into clinical practice.

Update Triggers

This review should be updated upon the emergence of:

1. New regulatory frameworks or standards for the validation and clinical use of health digital twins.
2. Breakthroughs in generative AI and computational modeling that significantly reduce the complexity or cost of creating and maintaining DTs.
3. Large-scale clinical trials demonstrating the efficacy and cost-effectiveness of DT-guided interventions in major disease areas.
4. Novel solutions addressing the socio-ethical challenges, particularly data governance and patient consent.

Near-Term Work

• Standardized Data Integration: Develop frameworks for seamless, real-time integration of heterogeneous data from sources like EHRs, wearables, and imaging (Vallée et al., 2023). Focus on creating interoperable data pipelines that ensure data fidelity and security.
• Multi-Scale Modeling: Advance from single-organ to multi-organ and whole-body models. Research should focus on coupling physiological models across different biological scales (e.g., cellular to organ level) to capture complex systemic interactions (Acero et al., 2020).
• AI/ML Model Enhancement: Improve the predictive accuracy and interpretability of AI/ML algorithms used for model calibration and real-time updates. Explore hybrid models combining mechanistic principles with data-driven approaches.

Validation Needs

• Clinical Validation Frameworks: Establish rigorous, multi-stage validation protocols against clinical trial data, real-world evidence, and prospective studies. Validation must assess not just predictive accuracy but also clinical utility and safety (Sun et al., 2023).
• Uncertainty Quantification: Develop methods to systematically quantify and communicate uncertainty in digital twin predictions to clinicians, which is crucial for trust and adoption in high-stakes decisions.

Datasets and Benchmarks

• Public Datasets: Curate and release large-scale, longitudinal, multi-modal public datasets for specific diseases (e.g., cardiovascular disease) to serve as a common ground for model development and comparison (Acero et al., 2020).
• Performance Benchmarks: Define standardized benchmarks and metrics to evaluate the performance of different digital twin models in specific clinical tasks, such as treatment response prediction or surgical planning.

Translation Barriers

• Data Privacy and Security: Overcoming regulatory and ethical hurdles related to sensitive patient data is paramount. Robust anonymization and federated learning techniques are needed.
• Computational Cost: The high computational demand for running complex simulations is a major barrier. Research into model reduction techniques and efficient hardware utilization is essential (Fuller et al., 2020).
• Regulatory Approval: Lack of clear regulatory pathways (e.g., from the FDA/EMA) for software-as-a-medical-device (SaMD) based on digital twins hinders commercialization and clinical adoption.

High-Value Experiments

• Prospective Clinical Trials: Conduct prospective trials where a digital twin guides personalized treatment decisions for one cohort, compared against the standard of care for a control group. A prime example would be optimizing drug dosage or surgical approach in cardiology (Acero et al., 2020).
• Surgical Simulation Studies: Validate the efficacy of digital twins for pre-operative surgical planning and rehearsal by comparing surgical outcomes (e.g., time, complications) between twin-assisted and conventional planning methods.