This is the first in a six-part series exploring how AI is transforming medical research and treatment.
In front of me is a heart that beats and moves like a human organ—but it isn’t real. There’s no blood flowing through it, nor is it housed in a human body. This is a digital twin: a computer-generated replica used to test implantable cardiovascular devices like stents and prosthetic valves. Once proven safe, these devices can be used in real patients.
Adsilico, the company behind this innovation, hasn’t stopped at creating a single digital heart. Leveraging artificial intelligence and vast amounts of data, they’ve generated numerous virtual hearts that mimic a wide range of biological characteristics, such as age, weight, gender, blood pressure, health conditions, and even ethnic backgrounds.
These AI-generated hearts address gaps in clinical data by enabling device testing on diverse populations that are often underrepresented in traditional trials. “This allows us to capture the full diversity of patient anatomies and physiological responses, something not possible with conventional methods,” says Adsilico CEO Sheena Macpherson. “Using AI to enhance device testing leads to the development of more inclusive and safer devices.”
A Safer Future for Medical Devices
The need for safer devices is clear. In 2018, an investigation by the International Consortium of Investigative Journalists revealed that faulty medical devices caused 83,000 deaths and over 1.7 million injuries globally.
Macpherson hopes AI-powered digital twins will help reduce these numbers. Testing devices thoroughly in a clinical trial environment is often cost-prohibitive. However, with virtual models, manufacturers can conduct extensive tests before moving to human trials.
“You can simulate various conditions like high or low blood pressure or different disease progressions to see how a device performs,” explains Macpherson. “This level of detail isn’t achievable in traditional trials.”
Digital twins also allow researchers to study underrepresented patient groups. Clinical trials have historically centered on white men, leaving gaps in data for other populations. By incorporating diverse AI simulations, manufacturers can develop devices that work for everyone.
How Digital Twins Work
Adsilico’s digital twins are built using a mix of cardiovascular data and imaging from real MRI and CT scans, with patient consent. This data helps create highly accurate models of the heart and its interactions with medical devices.
The process involves creating a digital replica of the device, which is then tested in a virtual heart through AI simulations. These simulations can be replicated across thousands of digital hearts, far surpassing the sample size of human or animal trials.
The potential cost and time savings are immense. For example, drug manufacturer Sanofi aims to reduce the testing period by 20% while increasing the success rate of clinical trials.
Digital Twins in Drug Development
Sanofi uses digital twins to accelerate drug development in fields such as immunology, oncology, and rare diseases. By combining data from real patients with AI, the company creates virtual “patients” that are integrated into control and placebo groups during trials.
Their AI models also simulate drug behavior, predicting how a medication will be absorbed and metabolized across different patient profiles. These tools help identify potential outcomes and risks more quickly than traditional methods.
With a 90% failure rate for new drugs during clinical development, even a modest improvement in success rates could save millions. “An increase of just 10% in success rates could result in $100 million in savings,” says Matt Truppo, Sanofi’s global head of research platforms.
Challenges and the Road Ahead
Despite their promise, digital twins have limitations. Charlie Paterson, an associate partner at PA Consulting, notes that these models are only as good as the data they’re trained on. Legacy data collection methods and a lack of representation for marginalized populations could perpetuate biases in virtual recreations.
Sanofi is addressing these gaps by sourcing data from third parties, such as electronic health records and biobanks, to complement its internal datasets.
Meanwhile, Adsilico envisions a future where AI digital twins could replace animal testing entirely. “A virtual model of a human heart is far closer to reality than a heart from a dog, cow, sheep, or pig, which are currently used in implantable device studies,” says Macpherson.
As AI continues to advance, digital twins are poised to transform medical research, making drug discovery and device development faster, safer, and more inclusive.