Digital Twins in Rheumatology: From Concept to Clinical Potential
From Mihai Turcu-Stiolica
Every patient tells a story that unfolds over time. Disease activity changes, treatments are started and adjusted, laboratory results fluctuate, imaging findings evolve and each clinical encounter adds another piece to the picture. Increasingly, we also have access to multiple layers of information about the same patient. The challenge and opportunity is to bring these different pieces together to better understand and anticipate the course of an individual disease.
Could we build a digital counterpart of a patient that evolves alongside them? This is the idea behind the “digital twin”.
What exactly is a digital twin?
Although there is no universally accepted definition of a patient digital twin, the concept generally refers to a patient-specific digital representation that integrates multidimensional data and can support prediction or decision-making. The concept originates from engineering, where digital representations of physical systems are used to monitor, analyse and predict their behaviour. In healthcare, digital twins are now being explored as a potential tool for personalised and predictive medicine. [1,2]
A patient digital twin could integrate different types of data collected over time: clinical characteristics, laboratory results, imaging, treatment history and response, and potentially genetic or other molecular information. The aim is not simply to collect more data, but to bring these data together into a dynamic representation that can help us understand how an individual patient’s disease may evolve.
Why could this be relevant to rheumatology?
Rheumatology may be particularly well suited to this approach. Many rheumatic diseases are heterogeneous and evolve over long periods, while the amount and diversity of information available for each patient continues to grow.
A digital twin could potentially integrate these different sources of information into a single, patient-specific computational representation that evolves as new information becomes available. This could help identify patterns across complex longitudinal data that may be difficult to capture when considering individual data sources in isolation.
A recently proposed Rheumatic Digital Twin framework illustrates this idea by combining electronic health records, clinical notes, imaging and omics data into a computational representation of the rheumatic disease journey. The proposed framework could, for example, be used to forecast disease flares, treatment changes or future disease activity. Importantly, owever, it remains a proposed framework rather than a clinically validated tool. [3]
From prediction to simulation
One of the most intriguing possibilities is that a digital twin could eventually do more than represent the patient’s current state. It could potentially help model possible future scenarios.
Could we estimate how a patient’s disease might evolve under different treatment strategies? Could we identify patterns associated with treatment response or disease progression? In the longer term, could a digital twin allow clinicians to explore hypothetical scenarios before making a clinical decision?
For now, these applications remain largely a research opportunity rather than an established clinical reality. A 2024 scoping review identified 80 claimed patient digital twins, of which 98% were still in preclinical phases. This highlights the considerable gap between the concept and routine clinical implementation. [2]
What are the challenges?
The potential is promising, but several questions remain. Digital twins depend on high-quality and sufficiently comprehensive data. Clinical information may be incomplete, collected at different time points, or stored in systems that do not easily communicate with each other. Combining clinical, imaging and molecular data also raises important questions around privacy and security.
Most importantly, these models require rigorous validation. An accurate prediction does not necessarily translate into a better clinical decision or improved patient outcomes. Verification, validation and uncertainty quantification are therefore essential to establish the reliability and clinical usefulness of digital twins. [4]
Bridging the gap between a promising computational concept and a clinically useful tool will require not only technical development, but also prospective evaluation of whether these systems actually improve clinical care.
Looking ahead
Digital twins are still an emerging concept in rheumatology. Yet, as personalised medicine increasingly relies on integrating complex longitudinal data, they offer an interesting possibility for the future.
Rather than replacing clinical judgement, a digital twin could one day complement it, helping clinicians integrate information, explore possible disease trajectories and personalise decision-making.
The potential lies not in creating a virtual replacement for the patient or the physician, but in building a richer computational representation of the patient’s disease journey.
Perhaps the next step in personalised rheumatology is not simply having more data, but finding better ways to turn those data into knowledge about the individual patient.
References
- Kataria S, Ravindran V. Digital twins in rheumatology: current status and potential. Rheumatology. 2026;65(6):keag260. doi:10.1093/rheumatology/keag260.
- Drummond D, Gonsard A, et al. Definitions and Characteristics of Patient Digital Twins Being Developed for Clinical Use: Scoping Review. J Med Internet Res. 2024;26:e58504.
doi:10.2196/58504. - Selani D, Knevel R, Reinders M, van den Akker EB. Rheumatic Digital Twin: Proposed Machine Learning–Based Multimodal Framework to Inform Clinical Decision-Making. J Med Internet Res. 2026;28:e86763. doi:10.2196/86763.
- Sel K, Hawkins-Daarud A, Chaudhuri A, et al. Survey and perspective on verification,
validation, and uncertainty quantification of digital twins for precision medicine. npj Digit Med. 2025;8:40. doi:10.1038/s41746-025-01447-y.