Also Read
By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. Cookie Policy.
Because of the complexity of the human body, very powerful computers will be needed to run calculations for reliable outputs.
In recent years, we are seeing a gradual integration of Artificial Intelligence in the healthcare system and especially in patient care. AI today helps doctors and clinicians with disease diagnosis, drug delivery, and even gives out health tips to optimise a patient's overall wellness. Recently, there has been a debate on the use of digital twin technology in healthcare. But what exactly is this?
As the name suggests, digital twin technology is nothing but the replication of a physical body or product into a digital presence. In healthcare, this means that a patient's body can be virtually replicated, on the basis of real time data. This data can be collected from sensors or medical reports and records. But how will this help doctors and patients?
To learn more about the use of digital twin technology in the field of medicine, we reached out to Prof Peter Coveney, Professor of Physical Chemistry, Director, Centre for Computational Science, University of College London. He shared his exclusive views on the subject in the following interview with TheHealthSite.com.
There are multiple benefits accruing from having digital twins in healthcare which includes:
The implementation of this technology is not without its own set of challenges. Here are some issues that have to be overcome.
What potential do digital twins hold for revolutionizing personalized medicine and enhancing patient care in the coming years?
The potential is indicated by the benefits I listed above. At the moment, patients are treated on the basis of the notion of the experience of doctors in treating past patients and making therapeutic decisions on the back of the apparent similarity of the current patient to ones of a 'similar' kind previously treated. This is prone to failure as no two patients are alike in detail. What works for one has no guarantee of working for another.
Being able to track and predict the long-term state of wellness of a person and keeping them in that state will avoid overburdening healthcare systems with aging and infirm individuals. With early warning of a person deviating from a state of wellness, pre-emptive action can be taken to mitigate this and return them to a healthy condition.
During the recent GREAT Talk at the British Council, the discussion rolled around how at a fundamental level, digital twin technology is about understanding how and why pathologies arise and how they can be treated. Medicine today of all sorts lacks mechanistic understanding of the kind that science requires for adequate explanations of such processes. That will come from using digital twins, because it brings computer models and simulations of the human body into conformance with observations of the state of health of individuals. These descriptions must be formulated mathematically and then coded/programmed into computers.
Because of the complexity of the human body, we need very powerful computers in order to run these calculations, especially if we are to glean actionable outcomes from them which depend on being able to certify their output as being reliable. That means reliable and trustworthy predictions which medics and clinicians can use to make decisions, some of which will be matters of life or death.
Currently, doctors are not trained in any of these ways of studying human processes. So, to advance in this domain, it is necessary for them to work very closely with groups with appropriate backgrounds in the physical and engineering sciences, as well as computer science and mathematics. My expertise lies in these domains, and this is how I can help to advance and shape the future of healthcare using digital twin technology. One part of this is accelerating drug discovery and making it much more effective than it is today this needs to be done by using computers for the pre-clinical steps, and also in the use of in silico clinical trials.
This can be done by the use of 'biomarkers', biological 'fingerprints' that reflect the state of wellness of an individual. There will typically be several of these for any disease condition as it evolves. Monitoring these for deviations from the state of wellness triggers the need to take further actions, which amounts to the application of a patient specific treatment on the basis of an individual's developing condition. The data acquired from the patient's current state can be fed directly into the digital twin and predictions run of what will change in the patient's state of wellness. Those predictions can then be used to advise clinicians of the actions necessary to attempt to mitigate these nascent changes and to return the patient to an acceptable state of wellness.
From my foregoing points (especially the answer to the third question above), it is clear that healthcare providers must participate very actively in collaborations with researchers to enable the latter group to identify, select and target a given disease case or cases. The challenges of modelling the complexity of human physiology and pathology are so great that international collaborations are essential to advance the field most effectively and quickly. Patient involvement is essential, even now, in preparing for this kind of future, because it self-evidently requires both their understanding of the approach and their direct participation, since each individual must provide their own data for personalised treatment. Much of that interaction can occur using mobile phones: India is in a unique position worldwide to participate in this new form of medical treatment, since everyone in the country has a mobile phone.
HPC and AI are and will continue to be essential to advance digital twin technology, as we need to obtain accurate and reliable predictions using high fidelity digital representations of individual patients. AI cannot be used in isolation as it makes predictions about a patient using other people's data and that is not reliable. But it has a powerful role to perform in accelerating much slower mechanistic models. Such technologies are important for population models as well for communicable/infectious diseases, we need to model transmission mechanisms and rates between large groups of individuals.
More excitingly, perhaps, populations of human digital twins can be run in computers in order to perform in silico clinical trials. For example, to determine the toxicity of newly developed drugs, we can perform an in silico trial of an assembly of tens to hundreds of human hearts, each one a person's own heart and thus different in detail from any other in the set of hearts being studied. We administer the drug into the model and study what impact it has on the way the hearts beat in a supercomputer, all the hearts being simulated concurrently (i.e. at the same time). This way, the population can be selected with a whole range of people - men, women, and various sub-groups within them, which is fully representative of the population we wish to apply such a new drug to.
From examples performed to date, it is clear that we see very detailed and specific responses by individual patients' hearts in the computer simulations; some may develop heart arrythmias that could lead to cardiac arrest and even death. But now, we are doing these studies the individual's own avatars, not their real selves with many manifest benefits.
Last word: this is the one and only way to develop medicine into a reliable scientific subject which fully accounts for the variability between individuals and reveals how to treat each and every person on an equal footing, based on their own data rather than other people's.