AI in Clinical Trials: Driving Innovation and Enhancing Processes

3-minute read

Artificial intelligence (AI) has a critical role to play in the management of data, and can dramatically accelerate the drug development and clinical research process more generally. The use of AI was a focus of several of the sessions at NEXT London this year, highlighting how valuable it can be for researchers. In one of the opening sessions, we highlighted how AI is already “turning data into insights”, benefitting researchers and, ultimately, patients. AI has significant potential to revolutionize the way clinical research is carried out, and we’re now beginning to see the consistent implementation of AI tools within trial processes.

One specific area of data collection within clinical research where AI may have particular benefits is imaging. In a session evaluating the role of AI in image processing and its ability to drive innovation, discussion centered on the overlap between research teams and AI technology and how it might solve some of the challenges emerging during image collection in clinical trials. Issues highlighted included the uploading of incorrect or incomplete images, as well as the increased volume of images involved in trials—which has only been exacerbated with the expansion of decentralization, allowing patients to take and submit their own pictures in some trials. During this session, Aleksandr Filippov, a Clinical Imaging Scientist at AstraZeneca, noted the importance of reducing the burden on trial sites and research teams when it comes to imaging and processing the associated data. Developments in imaging technologies have resulted in greater amounts of data. But processing, reviewing, and quality-controlling all the images and variations can be extremely time-consuming. AI tools can flag discrepancies during quality control and reduce the amount of time spent on this by researchers. These tools are only beneficial to researchers if the technology is reliable. Filippov explains that, “if we have a poor AI, then we’ve actually increased the burden” on those reviewing the data.

Regulating and overseeing AI to make sure that the tools researchers use are safe and effective has been a real challenge and, amid “all of this excitement and opportunity, there comes anxiety as well” according to Medidata’s Senior Director of Global Compliance and Strategy, Fiona Maini. In a session addressing recent regulatory developments with Maini, Parexel’s Dan Ballard highlighted the “super dynamic environment” that is the AI world at the moment. This has been reflected in the regulatory environment as well, with regulatory bodies trying to catch up with the rate of change in AI. Ballard noted how currently the industry is waiting for guidance, while regulators are waiting for input from the industry. Recent developments in regulation were noted in this session, highlighting the emergence of the trailblazing EU AI Act and, for the life sciences space, an EMA reflection paper which was published last year. But areas where progress is required were noted during this session, such as the need for harmonized definitions of key terms, including of ‘artificial intelligence’ itself. Meanwhile, the emergence of more generative AI capabilities has led to new challenges and Ballard suggested that a “GDPR-type regulation” may emerge in the US.

Despite this, generative AI has created a lot of excitement within the industry and is increasingly offering new innovative solutions to clinical researchers. This year’s Innovation Theatre hosted a session from Senior Director at Medidata AI, Tanmay Jain, on the implementation of generative AI to produce synthetic data for use within clinical trials. Jain noted significant recent investment in the synthetic data space which has coincided with increased actions from regulators who are starting to develop working groups on this technology as it becomes more prevalent in the industry. Jain highlighted the benefits of synthetic data which is much cheaper to generate than real data which requires experimentation. Researchers can significantly increase a trial sample size through the creation of synthetic patients with generative AI based on old datasets. Medidata’s experience across 25 years means it has access to vast amounts of data which offers the opportunity to embrace synthetic data strategies. One of the greatest benefits offered by synthetic data is the ability to anonymize data in order to protect patient privacy. As there’s no way to detect which data has been used to create synthetic patients, patient’s data remains diligently protected. This allows greater sharing of this non-attributable data by researchers accelerating the research process, and ultimately benefiting patients.