The emergence of a concept
First introduced at the University of Michigan in 2002 during a product lifecycle management course, the digital twin concept encompasses the notions of real space, virtual space, and a bidirectional flow of information between the two. The first use of the term ‘digital twin’ dates back to 2012 in a report by the National Aeronautics and Space Administration (NASA). The digital twin was defined as:
‘An integrated multiphysics, multiscale, probabilistic simulation of an as-built vehicle or system that uses the best available physical models, sensor updates, fleet history, etc., to mirror the life of its corresponding flying twin.’
The concept of digital twins (DTs) integrates advances in both cyber–physical systems and simulation, attracting interest from many fields. Beyond the manufacturing and aerospace industries, DTs are currently being developed for the management of cities, transport, agriculture, hospital management, design, and care coordination.
This growing interest can be attributed to developments in the rapid collection, storage, and sharing of data, together with computers being able to apply complex models and algorithms within a short period of time. DTs have recently emerged in several areas of healthcare, including precision medicine, clinical trials, and public health.
Their application has become increasingly apparent as they may serve as a tool to understand and simulate complex physiological processes, while also reducing the need for animal experimentation, which is estimated to involve 200 million animals each year. DTs can enable the direct translation of in vitro measurements into expected in vivo outcomes, whether in digital animal models or humans simulations.
What is a healthcare digital twin?
The general definition of a healthcare DT is a virtual replica of a real human patient through which clinicians can gain valuable insights, optimise treatment strategies, and deliver personalised care.
For specific healthcare domains, the operationalisation of this definition depends greatly on the underlying methodology and data used to construct the DT. Although the overall aim aligns with the definition above, a ‘cardiac’ DT differs considerably from a ‘drug response’ DT in methodology, data types, and implementation.
DTs can virtually represent any biological entity, from individual cells and cell cultures to tissue samples, entire organs, animal models, and ultimately full human representations. The initial state of a biological entity is used to initialise the corresponding DT, which can then simulate the results of real experiments by applying advanced computational and algorithmic techniques.
Healthcare DTs require large amounts of data, often from multiple sources and in various formats. These include measurements collected through smartphones or smartwatches, such as heart rate, temperature, and location, as well as data gathered at home, including blood pressure and blood oxygen saturation. They also include medical imaging data from CT and MRI scans, electrophysiology data, and various types of omics data collected through techniques such as sequencing, immunoprecipitation, and mass spectrometry.
How does a digital twin work?
There is a bidirectional connection between patients and their DTs: information flows from the patient to their virtual representation in order to simulate current and future states, while information flows back from the DT to the patient to support medical decision making.
The information flow from a DT back to the patient is determined by its predictions or simulations, which are model outputs specific to an individual patient. For example, if a DT predicts an adverse event, additional health monitoring or intervention can be initiated in a timely manner to manage the issue effectively.
Ideally, DTs should be indistinguishable from real patients in terms of observed characteristics, including monitored clinical variables and disease prognoses.
A few case studies
In diabetes management, the artificial pancreas is a prime example of a DT-like system that can significantly improve the monitoring and prediction of blood glucose levels, while also administering insulin based on non-invasive glucose monitoring methods.
Cardiovascular research has also seen many new developments in DT systems. Multiple types of data, including MRI and CT scans alongside electrophysiology measurements, can be integrated to create a digital heart model through which ECG patterns can be simulated and predicted in real time and across different locations. This capability enables the testing of patient-specific monitoring and treatment strategies and has the potential to significantly improve patient outcomes for patients with cardiovascular disease.
The approach generally consists of two distinct stages: anatomical twinning and functional twinning. Based on patient-specific CT or MRI scans, detailed 3D representations of the heart are generated. The second stage focuses on cardiac electrophysiology, where mathematical models are used to describe and simulate the electrical activity of the heart.
A fully comprehensive healthcare DT would also require the integration of different types of omics data. Certain models can incorporate single-cell RNA sequencing data and extracellular metabolite fluxes into DTs, providing insights into the metabolic phenotypes of cancer cells and enabling analysis of the effects of genetic alterations.
Using these data, gene-to-metabolic reaction links can be calculated and applied to simulate the effects of individual gene deletions. This approach offers a powerful tool for studying cancer phenotypes and identifying potential novel targets for therapeutic intervention.
Several challenges remain
Compared with DTs of engineered systems such as aircraft, several difficulties arise in the development of patient DTs. While an aircraft is created from components whose mechanical, thermal, chemical, and electrical properties are well understood, the physical and biological processes governing the human body remain only partially understood.
Similarly, humans design the assembly plan for an aircraft and therefore possess a multi-scale understanding of the system, from the smallest electrical unit to the entire structure. In contrast, cell-to-tissue, tissue-to-organ, and organ-to-organ interactions within the human body are still not fully understood.
While it is relatively straightforward to integrate a growing number of sensors into an aircraft to create an increasingly accurate digital replica, obtaining continuous data flows from the human body requires the implantation or wearing of sensors, bringing with it technical, medical, and ethical challenges.
Likewise, modifications to an aircraft can be implemented directly through actuators, whereas the inclusion of human factors – whether patients or healthcare professionals – introduces additional uncertainty into the system.
Unlike aircraft, whose ‘health’ depends solely on mechanical and physical factors, human health also includes emotional and spiritual dimensions that must be taken into account.
In addition, accurately replicating complex physiological systems requires a vast range of variables and parameters within DT models. Integrating these variables demands continued advances in computational power and modelling techniques.
Conclusion
Recent developments in DT research within healthcare hold considerable promise for simulating and understanding complex physiological processes, as well as supporting a wide range of medical applications.
Addressing challenges relating to data integration, computational power, and privacy will be crucial for advancing DT-based approaches. Overall, DTs have the potential to revolutionise medical care and precision medicine by offering increasingly personalised solutions for patients.
Véronique Ropion
Director of Business Strategy, Marketing & Corporate Communication, Pharmalys Ltd
Sources
- Bordukova M., Makarov N., Rodriguez-Esteban R., Schmich F., Menden M.P. (2024) Generative artificial intelligence empowers digital twins in drug discovery and clinical trials, Expert Opinion on Drug Discovery, 19:1, 33-42, DOI: 10.1080/17460441.2023.2273839
- Gonsard A, Genet M, Drummond D. Digital twins for chronic lung diseases. Eur Respir Rev 2024; 33: 240159 [DOI: 10.1183/16000617.0159-2024].
- Kamel Boulos, M.N.; Zhang, P. Digital Twins: From Personalised Medicine to Precision Public Health. J. Pers. Med. 2021, 11, 745. https://doi.org/10.3390/ jpm11080745
- Meijer, C.; Uh, H.-W.; el Bouhaddani, S. Digital Twins in Healthcare: Methodological Challenges and Opportunities. J. Pers. Med. 2023, 13, 1522. https://doi.org/ 10.3390/jpm13101522