If you’re into technology, you’ve probably heard the term “digital twins”. It originated in space engineering – at NASA – in the 2010s. Since then, groups from almost every industry have appropriated the term, including, most recently, healthcare.
What is a digital twin? At its most general, it’s a virtual copy of a physical object (jet engine), organism (person, animal) or process (circulation, drug metabolism), used for running simulations. Beyond that, the definition depends on who’s using the term. IBM’s “digital twins” of wind turbines are created using sensor data from real turbines, with two-way information flow between the twin and its real-life counterpart. Others claim the term for less accurate representations – a patient’s electronic health record might be referred to as their “digital twin”. Experts are debating a more accurate definition of this buzzword.
Meanwhile, nova’s approach to clinical trial simulation goes beyond digital twins – however they are eventually defined. We create virtual patients using mechanistic models of disease and clinical data. But we don’t just create one “twin”. The strength of our approach is its power to create multiple “twins” of a given individual – triplets, quadruplets and entire cohorts. Each copy can be used to simulate the impact of a different variable on the clinical outcome for a given person – be it age, blood pressure or concurrent medications.
Using ‘digital cohorts’ results in a deeper understanding of the impact of a medicine or trial design on clinical outcomes than any real-life trial could provide. It also enables more powerful synthetic control arms – not just capturing an individual’s identical twin taking placebo or a comparator drug, but including other identical siblings, each on a different regimen or with a different underlying health state. Digital cohorts, being unlimited in size and scope, can also help uncover hidden variables that impact treatment response – variables that a single ‘twin’ is unlikely to reveal. The result is a more comprehensive understanding of how drugs work and their effect on patients, before they are tested in people.
The growing use and discussion of the term ‘digital twin’ across healthcare reflects increased acceptance of modeling and simulation in R&D. As technology and data analysis mature, policymakers and regulators are incorporating these new methods in their guidelines. FDA is investigating Model Informed Drug Discovery approaches; in early October 2022, the US Senate passed the ‘Alternatives to Animal Testing’ bill, paving the way for in silico, in vitro and other technologies to gradually replace all or some animal testing in pre-clincal toxicology.
Progress in supplementing (or replacing) traditional R&D methods with potentially more powerful, efficient and accurate technology-backed approaches is, naturally, generating new vocabulary. We have ‘synthetic’ or ‘external’ control arms, ‘decentralized’ trials, ‘virtual’ patients, ‘digital twins’, and, at nova, ‘knowledge-based models’ and, yes, ‘digital cohorts’.
The precise definitions of these terms matter less than continuing to work on the underlying technologies to improve drug R&D and accelerate patient access to new medicines. If debating definitions helps raise awareness and acceptance of these approaches, we’re all for it.