Medical digital twins are rapidly gaining momentum. Until recently, however, the field lacked a clear definition and operational framework. A new health policy paper published in The Lancet Digital Health (Sadée et al., July 2025) addresses this gap. It offers a structured approach to building and deploying digital twins in clinical practice, paving the way for more precise and personalized care.
What Are Medical Digital Twins, and Why Do They Matter in Healthcare?
A medical digital twin is more than just a sophisticated model. It’s a dynamic, evolving virtual representation of an individual patient, continuously updated with multimodal health data—ranging from genomics and imaging to wearable device metrics. This “patient-in-silico” can simulate disease progression, predict treatment outcomes, and even suggest optimized interventions.
In simpler terms, a medical digital twin is a virtual replica of a patient, designed to evolve with real-time data. This approach marks a significant leap forward in personalized medicine and digital health innovation, offering clinicians a powerful tool for decision-making.
Five Pillars of the Medical Digital Twin
To bring clarity to the concept, the authors outline five essential components that make up a medical digital twin:
- Patient—the physical individual, or a specific organ/system of interest.
- Data Connection—multimodal data streams (e.g., EHRs, imaging, genomics, wearables) harmonized through AI-driven fusion.
- Patient-in-Silico—a high-fidelity model that simulates disease progression and treatment response.
- Interface—an AI-powered layer (e.g., large language models) enabling clinicians to interact with the twin.
- Twin Synchronization—continuous or episodic updates to reflect new patient data.
Together, these components form a living, learning system that supports real-time simulation, prediction, and clinical decision-making.
AI, Wearables, and Data Fusion: Technologies Powering Medical Digital Twins
Converging technologies fuel the rise of medical digital twins, each playing a crucial role:
- AI for feature extraction, data fusion, and predictive modeling.
- Mechanistic modeling for biological interpretability and extrapolation.
- Wearables and sequencing for continuous data acquisition.
- Physics-informed neural networks (PINNs) to merge AI with mechanistic insights.
Importantly, the fusion of AI and mechanistic models offers a promising path forward. By combining predictive power with biological insight, this hybrid approach helps overcome the limitations of each method when used in isolation. As a result, clinicians can build robust, personalized models that evolve alongside the patient.
Medical Digital Twin Use Cases: Oncology, Diabetes, and Beyond
Medical digital twins are already making an impact in clinical settings. For example:
- Oncology: Adaptive therapy trials use patient-specific tumor dynamics to optimize treatment schedules, improving progression-free survival.
- Diabetes: The ADVICE4U trial used an AI-based decision support system to guide insulin therapy. Acting as a simplified digital twin, the system provided personalized dosing recommendations, demonstrating non-inferiority to physician-led care and improving accessibility.
These examples highlight the potential of digital twins to transform care across diverse medical domains.
From Theory to Practice: Validating Medical Digital Twins
To move from concept to clinical reality, the authors propose retrospective data staging. This method validates digital twins using historical patient data, updated incrementally to simulate real-time evolution. It allows for rigorous testing without exposing patients to risk.
Additionally, the paper explores integrating digital twins with large language models like ChatGPT. In this setup, the patient-in-silico acts as a verified backend to conversational AI. While challenges such as hallucinations remain, this hybrid model could improve usability and build trust among clinicians.
Source: Sadée C, Testa S, Barba T, et al. Medical digital twins: enabling precision medicine and medical artificial intelligence. The Lancet Digital Health. July 2025; Vol 7: 100864.
