The virtual patients revolutionizing drug manufacturing

Posted: November 15, 2024

By now, you’re probably familiar with the concept of digital twins for industrial processes. But some companies and researchers are taking it a step further and using the technology to revolutionize drug and medical tech manufacturing by using digital twins in clinical trials.

Known as in-silico trials, using virtual patients allows companies to test novel drug candidates to assess safety and efficacy before human trials and even replace humans in some areas. In-silico trials have the potential to reduce the number of patients needed in clinical trials, and can speed up the process, as recruiting patients can take time. Importantly, in-silico trials can also help provide representation for patients that are often under-represented in traditional clinical trials, for example in cases of rare or childhood diseases, or women.

While digital twins won’t eliminate the need for human trials any time soon, they can help trials that do get to human stages get there faster, and could make them safer and more successful.

How does a digital twin of a patient work?

Clinical trials test the effect of medical treatments on patients. They often include a control arm, where a patient receives a placebo instead of the real drug. Researchers are trialing using digital twins of patients to provide this control arm in clinical trials.

In the clinical trial setting, a digital twin is a digital representation of either a full patient or just a specific organ or cells. The digital twin is built with measurements from, for example, a real heart, and returns predictions of how the treatment or medicine will affect the patient.

Scientists are also building digital twins of real patients, integrating data on disease biology, pathophysiology (the physiology associated with disease) and pharmacology into a digital framework. When digital twins match characteristics of actual patients, they can predict individual patient trajectories, allowing medical professionals to make more informed decisions. For example, the technology can help to forecast when patients are likely to feel better or worse, likely to have a health problem, or when their medication is or isn’t working.

Clinical trial digital twins tend to be built with mechanistic modeling or AI. Mechanistic modeling relies heavily on existing data, applying biology and physiology knowledge to create accurate simulations by predicting causal relationships. A mechanistic model will use information about the mechanisms by which a drug works to predict how a patient will react to it. AI and machine learning models can process a wider variety of data, are more scalable, and can produce synthetic data samples that are statistically similar to real, gathered data. AI models can predict outcomes for a group of people based on existing health outcome data, which is why one potential use is as the control arm in clinical trials.

How can digital twins be used in clinical trials?

Simulating control arms of trials

Virtual clinical trials mean that simultaneous testing of both a drug and a placebo on the same patients is possible, helping to solve the ethical issues of giving sick patients placebo drugs and reducing recruitment issues of patients not wanting to risk getting a placebo.

Phesi, a clinical development analytics business, built a digital twin that replicated patients receiving the standard line of treatment for chronic graft-versus-host disease, a complication that some patients face after a stem cell or bone marrow transplant. Phesi used a large database of real-world clinical trial data to build its digital twin, and performed an efficacy assessment of the standard drug, producing results that matched current literature.

In the future, Phesi hopes that it will be able to use this digital twin to simulate what patient outcomes are likely to be with the standard treatment so that it can compare the simulated outcomes with a patient’s actual outcomes from receiving a new trial treatment. Dr Gen Li, President of Phesi, said: “We have demonstrated that digital twins offer real potential to replace standard-of-care comparator arms to streamline the implementation of clinical trials and dramatically reduce patient burden.”

Research and development

Sanofi, a global healthcare and pharmaceutical company, is using digital twin technology in its research and development process to speed up drug development. The company uses quantitative systems pharmacology to build a digital twin of human patients, including data on disease biology and pharmacology to look at how potential medicines impact specific disease drivers, and the effects patients feel. With this framework, Sanofi can test how medicines work on specific disease pathways to establish the impact a medicine has on the disease and on the patient. For example, in an asthma study, Sanofi examined how a drug might affect proteins that control inflammation and evaluated the drug’s effect on lung function, such as exhalation rate.

Virtual patients allow Sanofi to test novel drug candidates at different early phases during development, allowing it to carry out faster and more accurate first assessments of the safety and efficacy of drug candidates.

Checking correct medicine dosage

Multinational pharmaceutical company Bayer has developed a virtual diabetes twin to predict levels of blood glucose and inform insulin dosing. It has also run simulations to inform dose selection for an anticoagulant medication, which allowed the company to confirm its planned doses were adequate.

It has shared its digital twin technology on an open-source platform, aiming to share knowledge and increase regulatory acceptance of the technology.

Predicting outcomes for specific patients

The Turing Research and Innovation Cluster in Digital Twins is pioneering digital twins research and innovation to address society’s most urgent challenges. At the cluster, researchers are building patient-specific models of hearts for patients with pulmonary arterial hypertension, a life-threatening heart disease. The project is designing and building fully accurate copies of patients’ hearts using medical records, hospital scans and data from wearable and implanted monitors, which will provide continuous, real-time data updates from the patients.

The research team—including engineers, clinicians, computational statisticians and research engineers—hope that the digital twin hearts will allow them to accurately track changes to each patient’s disease progression, their responses to treatment, and will enable personalized predictions. The project will also assess whether using digital twin heart models is feasible, scalable and affordable in NHS patient care pathways.

Catching device limitations earlier

At the University of Leeds, scientist Professor Alejandro Frangi is working on re-enacting a clinical trial for a prosthetic heart valve device virtually through his project INSILICO. The initial clinical trial was based on a randomized controlled trial design, which took nine years and cost millions. The project team are building a digital representation of the prosthetic heart valve and will test it on virtual patients using machine learning. INSILICO is hoping to replicate the original trial results and establish whether an in-silico trial could have predicted patient outcomes and limitations of the device quicker and more cheaply.

In a press release, Professor Frangi said, “The ultimate aim of this project is to anticipate virtually when a device is unlikely to work. Doing the testing at the beginning of the process will reduce the research and development phase from years to months and significantly reduce risks and welfare issues. The need for human confirmatory trials is unlikely to change soon. However, by the time we go to patients, we could have greater certainty of success, have identified as many failure modes as possible, and optimize the designs more thoroughly than today.”

What’s the future for digital twins in clinical trials?

Currently, the key barriers to digital twins in the healthcare space are quality and accessibility of data, and regulatory approvals.

For all digital twins, success relies on the quality of the data they receive. Building digital patients requires comprehensive existing data, limiting the diseases that digital twins can currently help with. In the future, algorithms that could improve deep learning and anticipate the unknowns in more complex diseases would mean that virtual patients could help improve drug development for more conditions.

As with any new technology, users and regulators need reassurance that they can rely on results from digital twins. At the moment, new medical devices and drugs go through extensive testing via lab-based experiments, animal studies and clinical trials. Now, regulatory bodies are starting to consider safety and efficacy of medical products obtained from computer modeling and simulation as it becomes a growing trend: As of March 2023, at least 14 drugs fully generated by AI have entered clinical trials. The Alan Turing Institute is collaborating with the FDA, which approves medical devices in the US, to develop processes for assessing the credibility of virtual clinical trials for medical devices. The FDA and European Medicines Agency (EMA) have also taken initiatives to support the integration of in-silico approaches to control arms.

With improved regulation and our ever-increasing volume of data, digital twins in clinical trials have the potential to completely change the way we conduct human trials. Digital twins are already paving the way for faster, safer and more effective medical interventions and offer hope for more personalized medicine. In the future, they could even change the way we think about medicine completely, offering the possibility of a shift to preventative rather than curative medicine.