Virtual Bodies for Real Drugs: AI May Speed Up Clinical Trials and Save More Lives

Artificial intelligence technology is helping researchers and clinicians test the safety and efficacy of a new drug without any impact on humans and animals.

By Pragati Verma, Contributor

For years, drug companies had only one way to find out if a new medicine or vaccine works: patient-based clinical trials. But a community of virtual patients is promising to modernize the drug testing process by predicting drugs’ effects.

The resulting studies, called in silico trials, use artificial intelligence (AI)-based simulation technology to mimic human physiology that pharmaceutical companies can use as a proxy for patients. Researchers and clinicians feed drug data into the simulated model to predict the drug’s exposure in patients, and, thereby, test the safety and efficacy of a drug without any consequences on humans and animals.

“The goal is to enable drug companies to [set up] the fastest and smallest clinical trials [required to] prove to the regulators that their drug is safe and effective,” explains William F Feehery, CEO of Certara, a company that performs in silico trials for pharma and biotech companies.

The company’s simulator—called Simcyp—leverages data collected from scientific studies to simulate the biological process and then uses drug data to model its effect on patients. Several other companies offer similar services, such as France-based Novadiscovery. Its software-as-a-service platform Jinko combines mathematical models of diseases and treatments with virtual patients, enabling drug companies to directly conduct their own in silico trials. InSilicoTrials, a spin-off of Italian company Promeditec, offers a web-based platform that uses simulations to test new drugs and medical devices in virtual patients.

The Promise of In Silico

Feehery says the Certara model has learned by ingesting data on more than 90 drugs and developing 20 different virtual populations, working with several pharmaceuticals for more than 20 years. Since 2014, Feehery claims to have supported more than 90 percent of novel drug approvals by the U.S. Food and Drug Administration (FDA).

“Pharma companies, as well as regulators, are embracing the technology,” he says, explaining that in silico trials can be executed quickly and for a fraction of the cost of a full-scale live trial. “A number of studies have shown how simulation models optimize cost and time spent on clinical studies,” he says. Extensive use of modeling and simulation software can reduce the cost of conducting clinical trials by $400 million to $1 billion for each drug, according to a study published in Applied Clinical Trials that evaluated three drugs—Aristida, Keytruda, and Cotellic.

In silico trials may also potentially protect public health. “Clinical trials are the most expensive part of developing a drug,” continues Feehery, “but there can be a big human cost, too. What if we give people enrolled in a trial a harmful drug or if the dosage is wrong?” he asks. The software modeling, he adds, “can be used to predict the safe and effective dosage before starting trials [on humans]. So you don’t need to start a trial by guessing dosage and then figuring out if you are in the range. We can use what we know about biology to make informed predictions and then verify it in the trial.”

Another challenge facing drug makers is to state with confidence how their drug might interact with any other drugs the patient takes. “It’s very hard to do a clinical trial testing the new drug’s interaction with every other drug that might be out there,” Feehery explains.

That’s where their simulator comes in, says Karen Rowland Yeo, senior vice president of client and regulatory strategy at Certara. She explains how they helped expedite the development and regulatory approvals of the oncology drug Imbruvica. “Our simulator predicted how Imbruvica would interact with 24 other drugs and what kind of dosage adjustment will be required if patients were given multiple drugs,” she says. “The drug was approved by FDA after two clinical trials when typically they would expect seven to eight trials.”

The technology can also help when it’s not easy to find and recruit patients for trials. Consider Swiss dermatology company Galderma’s topical acne cream Aklief. Galderma used Certera’s simulation to predict how Aklief would interact with other drugs and also provide dosing guidance for pediatric patients ages 9 to 17. “The FDA accepted our simulation in lieu of clinical studies [on real patients],” Yeo says. Galderma didn’t just cut time and cost to market, according to Yeo. “They could overcome the ethical and operational challenges of pediatric drug trials. Our simulator also helped avoid unnecessary drug exposure to patients and healthy volunteers,” she adds.

Speeding Up COVID-19 Treatment

After supporting drugs of more than 1,600 companies, Feehery is now hoping to harness their simulator to speed up trials for new vaccines and treatments for COVID-19. “Modeling and simulation are important for accelerating drugs to market, especially in a pandemic situation,” says Feehery.

“Modeling and simulation are important for accelerating drugs to market, especially in a pandemic situation.”

—Dr. William F. Feehery, CEO, Certara

Certara’s research teams are working on a number of programs to conduct virtual studies and use modeling and simulation to find out how safe and effective they can be for treating COVID-19. One such program is the COVID-19 Therapeutics Accelerator, a global initiative funded by Bill and Melinda Gates FoundationWellcome, and Mastercard to speed up the response to the pandemic by identifying, assessing, developing and scaling up treatments. Certara is also collaborating with AbbVieAcer TherapeuticsRed Hill Biopharma60 degrees Pharmaceuticals and DMTC (formerly the Defense Materials Technology Centre) in Australia to evaluate the preventative use of existing drugs such as Chloroquine as well as new therapies.

“Some of these drugs were not designed for COVID and might fail, but all of them will be much better off with some intelligent predictions about the right dosage and whether the drug being tested will be more effective in early-stage or intensive-care patients,” Feehery says.

Plugging AI Into Drug Discovery

As they investigate treatments for COVID-19, Certara researchers are working on ways to move their simulation model beyond clinical trials and plug it into the process of drug discovery itself with a platform for quantitative systems pharmacology.

The idea, according to Feehery, “is to use biosimulation capability to screen thousands and thousands of compounds to pick a drug” that is more likely to succeed in trials. “Simulators are working their way back in the drug design cycle. In the last four to five years, we have seen that companies are increasingly using them during drug development to make predictions about its effectiveness, even before they start thinking of clinical trials.”

The goal for the next decade is to tailor the model to each patient, with the option to predict a personalized dosage based on the patient’s physiological parameters. “It will be like developing a virtual twin of each patient to computationally identify and try the best medication even before actually treating a patient,” Yeo says.

If successful, AI-based biosimulation can transform the entire drug development process by selecting the right drug and right dosage for that specific patient early in treatment. “This could be a home run,” says Feehery. “This means more drugs could pass trials, succeed, and save more lives.”