The tipping point for digital twins in healthcare
Last month I presented for a wide audience of healthcare experts on the state of Digital therapeutics and explained how these different solutions (aka data silos) could be leveraged together in my personal digital twin.
Afterwards, the innovation architect of one of the leading Belgian payers (sickness insurance funds or mutualities) reached out to me and asked one simple question.
That question was not ‘what is a Digital Twin’? It was ‘should we build a Digital Twin for each of our 4.6 million members if we want to transform from a sick fund into a health fund? Which digital tools could we trust to guide us on this journey ?
That showed me the time is ripe for preparing the biggest transition our care system will ever experience: The transition from a sickcare system to a real healthcare system by default.
Now Is the perfect time to reflect on 2021 as the tipping points for digital twins in healthcare. People who follow me know the drill 🙂
Over 3000 years ago, in certain parts of China, doctors were paid to keep people healthy. Once you became ill, you no longer had to pay. This doctor could do so because he lived 24/7 amongst his villagers. Today, a doctor is 8755 hour a year disconnected from his/her patients. Now, enter technology. Enter the promise to replace these 8755 hours with trusted ways to collect, learn from and coach with data, from both patients today as well as healthy citizens tomorrow. The promise to be with the patient 24/7 not to cure, but to anticipate disease from striking.
With technology, we’ll see the birth of our virtual alter ego – our digital twin, or digital self, fed by data provided by the patient / citizen, but going beyond the medical records, classical physical and mental health parameters. Real real world data and any Social determinant of health will be a part of this picture. For additional context, I will refer to a recent blog here.
Incidentally, I was given the opportunity a few years ago to write the one pager description for the Health EU project, headed by the École polytechnique fédérale de Lausanne (EPFL). The aim was to raise over 1 Billion Euros to create a human Digital Twin for every European by 2030, capable of anticipating the 5 most common diseases. At that time, that vision was endorsed by 47 labs all across Europe, but for funding it was a few years too early.
The world is ready now. Here are some of the seminal papers and events from 2021 in the world of Digital Twin.
Growing investments in Digital Twin technologies (e.g. IoT, data streaming, 5G, machine learning) are introducing a new generation of intelligence with associated opportunities to help improve people’s lives. Accenture calls this the ‘Mirrored world’. Healthcare organisations develop Digital Twins to create living models of facilities, supply chains, medical products, as well as body parts and organs. Digital Twins provide answers to key questions that will soon be essential to every healthcare enterprise’s digital strategy. According to their report, a quarter – 25% – of healthcare executives indicate their organizations are experimenting with Digital Twins this year.
The leading technology newsletter VentureBeat published a comprehensive review article listing 21 ways medical Digital Twin applications promise to transform healthcare.
Additionally, this opinion piece in STAT news by Gopal Sarma argued on why we should consider simulating the entire U.S. health system. It refers to two relevant events. First, to the remarkable piece of internal infrastructure that Facebook (aka Meta) has released in the spring of 2020. Called Web-Enabled Simulation (WES), the platform is a detailed replica of Facebook, with artificial user accounts ranging from simple bots that browse the site to machine-learning-based agents that mimic social interactions. The sophistication of the platform is astonishing.
The second refers to a platform known as Archimedes, developed nearly 20 years ago at Kaiser Permanente with WES-like ambitions for the healthcare system. Built in collaboration with Sandia National Laboratories, Archimedes included models of patients, providers, interventions, policies, protocols, logistics, finances, and more. Moreover, the simulation was so granular that it modeled the relationships between a patient’s heart, lungs, kidneys, and other organs. The fact that such a simulation was possible before digitized health records, machine learning, and cloud computing suggests that its ambitions can certainly be revived today.
A full-scale simulation of the U.S. healthcare system would represent a profound accomplishment for the country. It would be a shared structure that enables coordination across the diversity of care delivery organizations, and would be a key resource for planning for – and preventing – future pandemics and biological attacks. Simulations can play critical roles as aggregators of data and models, and as computational “thought partners” for planning. In the case of the healthcare system, a full-scale simulation would make it possible to dynamically characterize the relationship between population health and supply chains, particularly in crisis situations.
On April 28, the world’s first academic Journal focusing on Digital Twins was published, showing the field becomes mature.
As a co-chair of the healthcare working group of the Digital Twin Consortium (DTC – chaired together with the fantastic Erin Bournival from Dell Technologies), I have decided to summarize a number of key events here.
One of the seminal Digital Twin papers this year, by a team from the Technical University of Dresden in Germany, discussed the use of Digital Twins of Multiple Sclerosis (MS) as a revolutionary tool to improve diagnosis, monitoring and therapy refining patients’ well-being, saving economic costs, and enabling prevention of disease progression.
Likewise, GNS Healthcare, an artificial intelligence company creating in silico patients that simulate drug treatment at the individual patient level, and Scipher Medicine, a precision immunology company matching patients with the most effective therapy, this year announced plans to develop and launch Gemini, the in silico Patient™ for rheumatoid arthritis (RA), to help patients reach target treatments sooner.
Gemini, The in silico Patient™ for RA, links drug treatment to patient characteristics to the complex genetic and molecular mechanisms and pathways driving clinical outcomes, enabling the simulation of disease progression and drug response for patients. Through these simulations, Gemini discovers novel drivers of progression and treatment response to enhance clinical trial design and comparative effectiveness evidence generation for the treatment of RA.
In an even more practical and timely use-case, computer power-house Dell has partnered with the i2b2 tranSMART foundation to create privacy-preserving digital twins to treat the long-haul symptoms of COVID-19 patients. The Long Covid program hopes to improve treatment for the 5% of COVID-19 patients who develop chronic health issues. The new tools integrate de-identified data (which refers to data from which all personally identifiable information has been removed), AI, and sophisticated models that allow researchers to perform millions of treatment simulations based on genetic background and medical history.
Moving further to clinical impact is Twin Health. Twin Health is building a platform to help people manage diabetes and other cardiometabolic conditions. Type-2 diabetes is considered one of “the diseases of civilization,” affecting more than 420 million people worldwide. With its incidence increasing at an alarming rate, there is more than ever a clear need for early detection solutions that can improve the ability to intervene.
Twin Health is including the concept of “digital twins” in its approach, the idea of collecting data about a person and using that to tailor their recommendations. The company uses data from continuous glucose monitors, fitness watches (wearables), and blood tests, along with consultations with healthcare providers and coaches. Its Digital Twin technology is intended to use this data to create a representation of each person’s metabolism. It sells its services to self-insured employers, health plans and health systems. With a recent investment of US$140 million, the company is currently conducting a randomized controlled trial of its platform to research methods of reducing patients’ blood glucose levels and their associated dependency on medications. Just before completingthis paper’s research, a separate study published data from 460 patients followed 90 days providing evidence that diabetes reversal progression occurs in both non-obese and obese patients and in patients with longer duration of diabetes. Additionally, the study identifies clinically-defined subgroups of patients where this reversal progression occurs more quickly, depending on baseline characteristics of the subgroups.
This research is similar to the European project Optomics. OPTOMICS (Combining optoacoustic imaging phenotypes and multi-omics to advance diabetes healthcare) is a 60-month project which was created in January 2021. The project was selected as one of the 149 proposals submitted for a call on Digital Twins for life sciences. OPTOMICS aims to develop a new methodology that will deliver a paradigm shift in type-2 diabetes healthcare.
It proposes an innovative approach to satisfy this need, by harnessing Artificial Intelligence and statistical methods combined with state-of-the-art multi-omics and optoacoustic imaging technology to model static and dynamic processes in type-2 diabetes evolution. This so-called Digital Twin model aims to improve prediction and early detection of individuals likely to develop the disease, which will improve the overall possibility for prevention. Simultaneously, the method will reveal potential risks for developing disease complications, all while personalizing patient treatment.
Meanwhile, University of Cambridge researchers propose using Digital Twins to predict individuals’ health issues over time and intervene early with personalized preventive care. They presented an interpretable Digital Twin model providing a holistic view over patients’ conditions. The end result is a graph depicting the digital twin with projected longitudinal changes in blood pressure and other vital signs. They tested their proof-of-concept on two clinical case studies combining information at organ, tissue, and cellular level showing the potential of their framework in clinical practice. When they homed in on cytokines, which affect blood pressure and have been associated with COVID-19 infection, the technique performed well in various case simulations. This graph modelling approach generates integrated predictions that translate into patients trajectories.
Finally, of note is the EU project SimCardioTest, a collaborative research project between 10 organizations from 6 European countries and the United States funded by the European Commission. Here they showcase how a better understanding of the design of new tools for predicting cardiac pathologies will accelerate the adoption of computer simulations for testing drugs and medical devices.
Similarly, we’ve seen the real kickoff in silico clinical trials with synthetic data. In-silico trials rely on virtual populations and interventions simulated using patient-specific models and may offer a solution to lower trial costs.
Researchers from Leeds (UK) and leuven (Belgium) have demonstrated that in-silico trials of endovascular medical devices can: (i) replicate findings of conventional clinical trials, and (ii) perform virtual experiments and sub-group analyses that are difficult or impossible in conventional trials to discover new insights on treatment failure, e.g. in the presence of side-branches or hypertension.
Companies offering such new ways of performing clinical trials like Unlearn.AI, MD.Clone and Novadiscovery came to the forefront this year striking deals with leading pharma companies adopting their technology.
A modular computational framework for medical digital twins
In my book ‘Your guide to delight’, I anticipated that medical Digital Twins would play an important role in medicine in the future, not only to treat sick patients better but to prevent them from becoming sick in the first place.
The key ingredients for enabling both goals are:
1) an ever improving – but never to be complete – understanding of human biology and the homeostatic mechanisms that help us maintain health;
2) ever-improving data that capture the various determinants of our health from genome, proteome, microbiome sequences, data from sensors, behavioral data, social determinants,… ;
3) the technologies that allow us to incorporate 1) and 2) into computational models, both mechanistic and data driven; and
4) the increasingly common collaboration between the clinical and the computational sciences needed to create sufficiently credible computational models that have value in the real world.
What is still missing is the technological infrastructure to combine these ingredients. Unlike the practice in industry, biological and biomedical research is conducted largely by individual investigators around the world collaborating with each other as needed and communicating extensively through conferences, publications, and social media. While this research structure maximizes creativity in research, it requires additional technological infrastructure to “crowdsource” the output of these individual efforts into a coherent whole, for instance, a model that combines all the determinants over the course of a viral infection for an individual patient.
The methodology presented in this great paper is an open-source model architecture that for the first time satisfies all the requirements for distributed model building. We believe that 2022 will welcome the first real examples hereof and we look forward to spreading such developments via the very same Digital Twin Consortium mentioned previously.
Which brings me to the last point: A few weeks ago, Europe has launched the Implementation Roadmap towards its grand Europe’s Beating Cancer Plan (EBCP).
Near the top of the major section listing the Actions to support cancer prevention and care through new cancer research and an innovative ecosystem, there is an entry for the generation of a repository of digital twins in healthcare supporting cancer treatment. Take that for starters:
From the above, I hope it became clear that there is no longer a need to explain what a Digital Twin is, even in a healthcare context. Incidentally, did you notice I omitted a definition at the beginning? Hence now is the time to start executing on its promise. We warmly welcome you to join the Digital Twin consortium and to co-create the future of Digital Twins in healthcare with us. Finally, at this site, we collect contemporary Digital Twin research to bring you up to speed and keep you up to date. I invite you to take a look. Talk soon. And not to my avatar.