Digital twins in healthcare and drug discovery: From idea to success stories

The term "digital twins" has its origins in engineering and computer science, where it refers to a virtual replica of a physical object or system. Simply put, it is a sophisticated, dynamic, real-time digital representation of a physical entity created using advanced computational models.

The sole purpose of these models is to simulate how an object or process would change under a specific scenario. The aerospace industry has been particularly successful in deploying digital twins, where real-time data is fed to predict how the plane will behave to improve safety and performance. Recently, this concept has gained traction in healthcare, owing to advancements in AI, availability of large biological datasets, and increased computing power. In life sciences, digital twins can represent patients, cells, organs, or non-biological systems, such as twins of health systems or a drug manufacturing process.

Digital twins have tremendous potential to revolutionise drug discovery and enable precision healthcare for patients. They can accurately model disease pathways, predict the efficacy of potential treatments, and can be used to forecast various elements of their patient heath and disease progression. These capabilities can significantly reduce the time and cost involved in drug development, as well as provide a more targeted approach to treatment delivery.

Digital twins vs. computation models

Life sciences has seen a plethora of biological computational models, which use mathematical equations and computational techniques to simulate biological processes like metabolic pathways and disease progression, over the years.

Though computational models have been very helpful in the past to simulate specific aspects of a system, it is very difficult to mechanistically model the trillions of cells and interactions between genes and proteins at a full body level. Digital twins go beyond traditional biological models by (1) creating a real-time simulation of a physical entity that is at a system level, (2) providing the ability to integrate real-time data and not be fixed by the mechanistic equations, and (3) enabling predictive capabilities to predict states of the system.

Building digital twins of human biological systems is a Herculean task. We are talking about modelling trillions of cells and describing an even higher number of pairwise interactions between thousands of genes and proteins at a cellular level. While having a digital twin of the whole body simulating every single aspect is not feasible, several companies have made strides in building digital twins simulating specific systems or applications in human biology.

Some examples of companies building digital twins for drug discovery and healthcare include:

Unlearn.AI

Founded: 2017 | HQ: San Francisco, California, USA
Data used: Patient-level clinical trial; electronic health data.
Application: Unlearn.AI generates digital twins of clinical trial participants to improve the efficiency and confidence of randomised clinical trials. These models provide comprehensive forecasts of clinical outcomes, reducing the time and cost associated with clinical research, with the aim of reducing the total number of patients required to be enrolled in the trial.

Twin Health

Founded: 2018 |HQ: Mountain View, California, USA
Data used: Metabolic data, lifestyle profiles, and health data collected from wearable devices.
Application: Twin Health uses digital twins to create personalised health management strategies for diabetes, providing customised recommendations for optimal health management and preventive care.

Predictiv

Founded: 2022 |HQ: Bedford, Massachusetts, USA
Data used: Genetic, health history, and lifestyle.
Application: Predictiv generates digital twins to predict disease risks and personalise healthcare interventions integrating latest genetic research continuously. These models help in early disease detection and tailored treatment planning, enhancing preventive care and patient outcomes.

Predisurge

Founded: 2017 |HQ: Lyon, France
Data used: Patient-specific cardiac data.
Application: Predisurge develops patient-specific digital twins to simulate the cardiovascular system's behaviour in arteries and cardiac valves, particularly useful in planning surgeries and other medical procedures for cardiovascular diseases.

Q.bio

Founded: 2015 | HQ: Redwood City, California, USA
Data used: Genetics, biochemistry, imaging, quantitative anatomical data, vitals, wearable data, medical, family & social history, and more.
Application: Q.bio creates digital twins to provide a holistic view of an individual's health that would allow forecast of future biological states in several diseases and anatomical problems.

Real-world examples and success stories

Of the above companies, Unlearn.AI has achieved significant regulatory milestones, including the European Medicines Agency's (EMA) draft qualification opinion on their Procova procedure, which integrates digital twins to enhance longitudinal clinical trial efficiency. They are currently deploying their platform with several pharma companies, including Merck KGaA and QurAlis, to improve power in clinical trials and reduce the number of patients required in the control arms for immunology drugs and ALS, respectively.

Meanwhile, Twin Health's technology has shown promising results in clinical studies, where 95% of the diabetes patients demonstrated reducing HbA1c levels and improved weight loss, highlighting the potential of digital twins to transform chronic disease management.

And Predisurge's digital twin technology has been validated through clinical studies, confirming its accuracy and effectiveness in simulating cardiovascular interventions. It is currently deployed at 50+ medical centres, having benefitted 500+ patients so far.

The future outlook

Digital twins represent a significant advancement over traditional biological models, offering a dynamic, holistic, and predictive approach to healthcare. By integrating real-time data and leveraging advanced computational techniques, digital twins have the potential to transform patient care, streamline drug discovery, and promote proactive healthcare.

The two parallel technologies that have been growing in related areas of biopharma and healthcare research are blockchain and synthetic data generation, which can offer high synergies to digital twins. Pharma companies collect extensive research data, patient health records, and clinical trial data, offering valuable insights. However, strict privacy regulations often restrict data sharing, due to re-identification risks. Synthetic data created using generative models to replicate the statistical properties of actual clinical or research data, while removing personal identifiers and shared over decentralised and immutable blockchain ledgers, can allow large amounts of data to be accessible for training of digital twin models.

As technology continues to evolve, the adoption of digital twins in healthcare will likely expand, paving the way for a new era of precision medicine and personalised treatment.