6 Digital Twin Applications in Healthcare – The Revolution of the Next Decade

Digital Twins Healtcare

Digital Twin is a broad concept that encompasses many technologies and applications. A digital twin is a digital replica of a physical object or an intangible system. However, these virtual copies are not a complete representation of the complex multi-physics nature of the entity, if not a digital twin capture a part of it.

For example, you can have a digital twin of the heart, which represents the electrical behavior of the organ for set up a pacemaker. Or a digital twin of the chemical aspect of the hearth for drug development, or the mechanical response for surgery simulation. A multiphysics digital twin can also be focused on monitoring a particular disease/event like predict a heart attack, or a digital twin intended for design a specific medical device such as the optimization of a vena cava filter.

The takeaway is – a digital twin is not a golden model that captures the complete nature of the patient, at least not with the current state-of-the-art, if not a model with a specific purpose and scope of application. That is how we can ensure its reliability.

Digital Twins will shift the current treatment selection based on the state of the patient of today to an optimized state of the patient of tomorrow [1] and it would be a key part of the P4-Medicine – predictive, preventive, personalized, and participative.

Applications of digital twins in healthcare

Applications of digital twins in healthcare

Each of us is so unique in many ways that most of the current treatments conceived for the “average patient” are highly inefficient to the “actual patient”. From ineffective medication and surgery dissatisfaction to device rejection/replacement. Digital twins will be an essential piece to move from one-treatment-fits-all to personalized medicine.

Here, I summarize the different applications in which the digital twin technology will improve healthcare.

1. Diagnosis and treatment decision support – software as a medical device

The patient’s digital twin feed from different health data sources like imaging records, in-person measurements, laboratory results, and genetic will assist during diagnosis. The patient model will simulate the health status of the patient as captured from available clinical data and infer the missing parameters from statistical models. For instance, the combination of cardiovascular imaging and computational fluid dynamics enables non-invasive characterizations of flow fields and the calculation of diagnostic metrics [1]. 

2. Patient monitoring – wearables

Smaller and more comfortable wearables (sensors) will be used to feed with real-time data our digital twin in the cloud. With enough understanding of disease progression and continuous patient data collection via heath trackers (biometric, behavioral, emotional, cognitive, psychosocial…) we can develop models that detect symptoms at early stages, given doctors and users the capacity to diagnose the patient before getting ill. Besides, during treatment, we will be able to evaluate if the treatment is being effective.

There are already many sources of data that can feed our digital twin, to tailor our risk factors like medical records, lab test results, pharmacy data, wellness, and disease management data, well-being device-generated data, and social determinants such as zip code, local weather, buying habits… [2].

3. Surgery simulation – surgery risk assessment

Surgery by definition is personalized. From the current state to the best outcome, the surgery is tailored to the patient’s needs. Personalization is critical to increase intervention success and reduce patient risk. Digital twins will help by simulating an invasive clinical procedure to predict the outcome before the therapy is selected [3]. From medical device selection (position, orientation, dimension…) to surgical variable determination (magnitude, angle, shape…).

4. Medical devices design and optimization – MedTech

Two realms converge here. On one hand, we have the patient’s digital twin that has the specific characteristic of the patient and on the other hand, the medical device’s digital twin that captures the device design. We can correlate both models to see what happens when a particular device is installed into a specific patient. This is the case of populations that cannot be investigated clinically without harm, such as in patients with rare diseases or pediatric patients [3].

Digital twins are also very useful in optimization tasks like the improvement of a device’s performance by running hundreds of simulations with different conditions and different patients. Besides, with the emergence of 3D printing technology, patient digital twins can lead to the personalization of medical devices by creating unique designs for each patient.

5. Drug development and dosage optimization – in silico clinical trials

We can treat computationally a digital twin with thousands of drugs in order to identify the best one or ones for that specific case. However, this does not need to stop at the drugs that already exist. We can create a digital cohort of real patients with different phenotypes, which share symptoms, and test new potential drugs to predict with one has more possibilities to success as well as the optimal dosage. Improving the first shoot will reduce the number of clinical trials necessary.

In silico clinical trial will shed light on processes that take years to be observed in vivo or assess the risk of rare cases where a randomized clinical study would need thousands of patients to observe just a few of such cases [4].

6. Regulatory decision making

Since 2016, both the US Congress and the European Parliaments started to include modeling and simulation among the sources of evidence in the regulatory process of biomedical products [4]. In particular, the FDA has committed to transform digital evidence into a valuable regulatory tool because of its potential for cost-savings in evaluating medical devices.

Besides, some companies have stated that the cost for clinical trials may soon outpace revenue, which will accelerate the shift of the industry towards other relevant and reliable data sources for demonstrating the safety and effectiveness of medical devices and pharmaceutical products [3].

Technologies behind digital twins

Technologies behind digital twins

As the application and purpose of a digital twin vary also the modeling methodology differs from one application to another.

The technologies behind digital twins can be summarized in two synergic groups. The induction approach, statistical models that learned from data, and the deduction approach, mechanistic models that integrate multiscale knowledge and data.

Digital twins will leverage both approaches to deliver accurate predictions of the underlying causes of disease and the pathways to sustain or restore health [1]. 

1. Statistical modeling – statistical inference and artificial intelligence

Statistical modeling includes all the mathematical methods that infer relationships from data like Bayesian and frequentist inference as well as artificial learning methods. Statistical models follow an inductive approach. They allow the extraction and optimal combination of individualized biomarkers with mathematical rules. For that reason, in some cases, they are considered black boxes. These methods are used more often on patient monitoring, diagnosis, and treatment decision support.

2. Mechanistic modeling – simulation

Mechanistic modeling encapsulates all simulation methods based on our knowledge of physiology and the fundamental laws of physics and chemistry, like solid mechanics, fluid dynamics, heat transfer, electromagnetism, acoustics, and optics. Mechanistic models follow a deductive approach. They provide a framework to integrate and augment experimental and clinical data, identifying mechanisms and predicting outcomes, even under unseen scenarios. For that reason, they are considered white boxes. These methods are used more often on surgery simulation, medical device optimization, drug development, and regulatory decision making.

3. Other technologies – medical imaging and wearables.

Advances in medical imaging and wearables will have a great impact on the development of digital twins in healthcare. Medical imaging tools help to capture the state of the patient, its anatomy and physiology, and are one of the main inputs for mechanistic models. Wearables will be key to capturing real-time patient data for patient monitoring and statistical models set up.

The digital twin of the future – vision for the next decade

The technology to create digital twins in healthcare exists today. The challenge is for companies and institutions to start testing and applying this technology to specific healthcare problems.

Within the healthcare industry, the most advanced digital twin applications are being developed in cardiology: both the scientific knowledge [1,5,6] and the industrials developments [Dassault SystemesSiemens HealthineersPhilipsHeartFlowinHeartFEops]. Application in other fields will follow as the technology matures.

There are still many challenges to overcome, but digital twins will be an important piece of the healthcare of the future.

References

[1]        Corral-Acero J, Margara F, Marciniak M, Rodero C, Loncaric F, Feng Y, et al. The ‘Digital Twin’ to enable the vision of precision cardiology. Eur Heart J 2020

[2]        Schwartz SM, Wildenhaus K, Bucher A, Byrd B. Digital Twins and the Emerging Science of Self: Implications for Digital Health Experience Design and “Small” Data. Front Comput Sci 2020

[3]        Morrison TM, Pathmanathan P, Adwan M, Margerrison E. Advancing regulatory science with computational modeling for medical devices at the FDA’s office of science and engineering laboratories. Front Med 2018

[4]        Pappalardo F, Russo G, Tshinanu FM, Viceconti M. In silico clinical trials: Concepts and early adoptions. Brief Bioinform 2019

[5]        Niederer SA, Lumens J, Trayanova NA. Computational models in cardiology. Nat Rev Cardiol 2019

[6]        Shameer K, Johnson KW, Glicksberg BS, Dudley JT, Sengupta PP. Machine learning in cardiovascular medicine: Are we there yet? Heart 2018

Author: Enrique Morales Orcajo, PhD

I am an engineer, scientist, and traveler based in Europe. I write about how to consume and digest scientific studies in a practical and efficient way. My goal is to help you make more sound scientific judgments.

Leave a Reply

Your email address will not be published.