How ‘digital duplicates' might accelerate pharmaceutical research
A researcher at Adsilico examines a virtual model of a heart
Artificial Intelligence can generate numerous models of a virtual heart
This marks the beginning of a six-part series exploring the transformation of medical research and treatments through Artificial Intelligence.
The heart before me pulsates and behaves as if it were a living human organ, though it circulates no blood within and exists outside of a human body.
This heart, a digitally created counterpart or ‘digital twin', serves to evaluate implantable cardiovascular mechanisms such as stents and prosthetic valves, which, upon verification of safety, are intended for later use in actual patients.
However, the developers behind the heart, Adsilico, have advanced beyond merely crafting a single precise model.
Leveraging artificial intelligence and vast data quantities, they've developed various distinct hearts.
The synthetic hearts produced by AI can mirror not only biological characteristics such as weight, age, gender, and blood pressure but also health statuses and ethnic origins.
Given these variances are frequently unrepresented in clinical data, digital twin hearts enable device makers to carry out trials among broader populations than possible with human trials, or those solely using digital twins devoid of AI.
“This enables the encompassing of the entire spectrum of patient anatomies and physiological reactions, unachievable through traditional approaches. Employing AI to improve device testing fosters the creation of appliances that are both more encompassing and secure,” states Adsilico CEO Sheena Macpherson.
A 2018 probe by the International Consortium of Investigative Journalists disclosed that medical devices were responsible for 83,000 fatalities and upwards of 1.7 million injuries.
Ms. Macpherson is optimistic that AI-driven digital twins will reduce those figures.
“To enhance the safety of these devices significantly, they must undergo more extensive testing, which is not practicable in a clinical trial setting because of the high costs involved,” explains Ms. Macpherson, from Northumberland.
“Hence, the intent is to employ the computer-simulated version to ensure that any procedure has been exhaustively verified prior to human testing,”
“A portion of those fatalities – along with the ensuing legal actions – might have been prevented through more comprehensive evaluation, which also yields more precise outcomes.”
“It's possible to utilize the identical [digital] heart to evaluate its performance under varying conditions of blood pressure or through diverse stages of disease to ascertain any impact on the device.”
Ms Macpherson notes, “Through [digital] testing, manufacturers of medical devices gain a wealth of insights. It also enables testing across diverse patient subgroups, moving beyond the traditional focus on white males characteristic of clinical trials.”
Woman scientist engages with a display
Adsilico's artificial intelligence models utilize training from a mix of cardiac data and authentic MRI and CT imaging, incorporating medical visuals from patients who have given consent.
The information is sourced from intricate anatomical configurations of the heart to facilitate the creation of precise digital models demonstrating how medical apparatuses will engage with various patient anatomies.
Adsilico's experiments entail crafting a digital counterpart of the device under evaluation, subsequently integrating it into a virtual heart within an AI-produced simulation.
This entire process occurs within a computer, allowing the experiment to be duplicated across thousands of hearts, each an AI-simulated replica of a real human heart. In contrast, human and animal trials usually include only hundreds of subjects.
Headline
Further Insights into AI and Healthcare
The greatest motivation for pharmaceutical and device makers to augment clinical trials with AI digital twins lies in the significant reduction in time required, leading to substantial cost savings as well.
For instance, pharmaceutical company Sanofi aims to shorten the testing duration by 20% and enhance the probability of success. It employs digital twin technology across its specializations in immunology, oncology, and rare diseases.
By utilizing biological data from real individuals, Sanofi generates AI-driven simulated patients—not exact replicas of specific persons—that can be distributed among both the control and placebo groups in the trial.
Sanofi's AI initiatives further produce computer-simulated models of the drug under investigation, emulating characteristics such as the drug's absorption throughout the body, allowing for tests on the AI patients. This system also forecasts their responses, mirroring the actual trial procedure.
Matt Truppo, the global head of research platforms at Sanofi, delivers a presentation
Photo credit: Sanofi
Photo description,
Utilizing digital twins could result in significant cost reductions for pharmaceutical companies, according to Matt Truppo
“Given the industry's 90% failure rate for new drugs in clinical development, just a 10% improvement in success rates through technologies like digital twins could lead to savings of $100m, considering the expensive nature of late-phase clinical trials,” remarks Matt Truppo, the global head of research platforms and computational research and development at Sanofi.
“The outcomes to date have been encouraging,” adds Mr. Truppo, who operates out of Boston, US.
“Much work remains to be done. The conditions we aim to influence today are immensely intricate. In this effort, technologies such as AI play a critical role. Enhancing the latest digital twins with precise AI representations of complicated human biology represents the forthcoming boundary.”
Charlie Paterson examines a whiteboard adorned with pink sticky notes
Source of image, PA Consulting
Photo description,
Charlie Paterson states that the effectiveness of AI hinges on the quality of its training data.
However, Charlie Paterson, an associate partner at PA Consulting and former NHS service manager, notes that digital twins could have flaws.
He emphasizes that the efficacy of these twins is entirely dependent on the quality of their training data.
“Given outdated data collection techniques and underrepresentation of marginalized groups, there's a risk that biases could still be perpetuated in the programming of virtual replicas of people.”
Sanofi acknowledges the challenge of using restricted historical data to educate its AI and is taking steps to address this issue.
To bridge deficiencies within its own databases, consisting of millions of data elements from the thousands of participants in its annual trials, the company acquires additional information from external sources, such as electronic medical records and biobanks.
At Adsilico, Ms. Macpherson remains optimistic that in the future, AI digital twin technology could potentially abolish the use of animal testing in clinical trials, which, as of now, is still viewed as a crucial phase in the examination of drugs and devices.
This News Story is brought to you by QLM Business News, your Digital Media Channel
For more business news stories also follow us on Facebook, X and Youtube.
Disclaimer: All images presented herein are intended solely for illustrative purposes and may not accurately depict the true likeness of the subjects, objects, or individuals referenced in the accompanying news stories.
