AI platform predicts clinical trial outcomes
Predictive analytics company Opyl has developed software that uses artificial intelligence (AI) to predict the probability of a vaccine, drug, diagnostic or medical device succeeding in clinical trials.
The uncertainty and delays around clinical trials remains a frustrating challenge and significant risk for biopharmaceutical companies, investors, clinical strategists and medical researchers, who would value the ability to predict the outcome of an individual clinical trial and have the opportunity to adjust the variable to improve its chances of success. Previous studies have demonstrated that on average only 13.8% of all drugs in phase 1 clinical trials eventually win approval from regulators and enter the market, with vaccines having one of the highest success rates (33.4%) and cancer drugs having the lowest (3%).
Knowing the predicted probability score, and having an opportunity before the trial commences to improve upon the trial design, would save hundreds of millions of dollars. The aim of the Opyl software platform is to work with drug and device development companies to refine their clinical trial approaches to improve the outcomes of their clinical studies, reducing costs and accelerating the timeline to get new treatments to patients.
The AI platform uses current and historical global data and considers factors including the numbers of participants in each trial; the dropout rate from those trials; how long each trial will take; the end point for each trial relative to related studies; and the mode of action, such as type of protein or vector being employed in a program. The platform has already delivered early results that are claimed to be more accurate than previously published models and with more functional features, including the potential to optimise trial design.
“Our approach is to use AI to not just predict the outcome, but to demonstrate that changing specific clinical trials variables can improve the probability of success,” said Opyl CEO Michelle Gallaher.
To illustrate a proof of concept, the company applied the platform to 475 clinical trials for vaccines and therapies targeting COVID-19. The results were as follows:
- Therapies show a much higher probability of success in clinical studies than vaccines.
- To date Opyl has identified the two vaccines most likely to succeed their current stage of development (phase) compared to all others.
- Antibody therapies have the best probability of success of getting a positive phase 3 outcome over all other programs.
The next stage of the platform’s development will involve increasing the data pool from additional clinical data sources and expanding the variables in order to further train the algorithm and refine the specificity and reliability.
“Although looking at the current pipeline of COVID-19 programs is an initial application of the AI platform, we are not limiting ourselves to just COVID-19 trials,” Gallaher stated. “The AI platform can be applied to all drugs, diagnostics, vaccines and medical devices about to begin or in clinical trials, and our goal is to improve the clinical trial process, which will in turn save money, time and ensure patients can access treatment options sooner.”
Opyl is now reaching out to those groups that may have an interest in the findings from the COVID-19 work, as well as continuing discussions with companies and partners on other applications of the technology.
Researchers have successfully used machine learning to complete cumbersome materials science...
Researchers have found two new mammogram-based measures for breast cancer risk, which could...
Using machine learning, researchers have identified distinct patterns of coordinated activity...