Artificial Intelligence (AI) techniques across many differentiating industries and domains have grown significantly over the last few years. As in other industries, AI offers great promise across the clinical trial landscape to address common challenges like recruiting and retaining clinical trial participants, identifying the right sites for a particular study, and predicting outcomes to improve the likelihood of a clinical trial’s success.
Looking ahead to 2024, technological advancements in AI, particularly with generative AI and Large-Language Models (LLMs), will continue to grow due to computing availability and the improvement of algorithmic techniques. These advancements in AI, coupled with technology, can address the many challenges and pain points across clinical trials.
According to the National Institutes of Health, traditional clinical trial design and execution are time-consuming and inefficient because of many manual processes, leading to only ~10% associated success rate of drugs that receive FDA approval. Given that the drug pipeline has more than 6,000 molecules in clinical development, sponsors require a massive amount of energy, resources, and funds to get these through all phases of trials. The current technology landscape provides the industry an opportunity to pivot to data and AI-driven solutions that are highly enabled by technology. This means not just digitizing a paper process but reimagining the clinical trial life cycle.
The key challenges in a clinical trial life cycle have remained the same for several years and will only continue to get more intractable if not addressed. Finding and keeping patients, determining optimal sites, data and document management, as well as detecting anomalies in data and the associated time to take corrective actions, will be further exacerbated if the replication of inefficient trial designs continues to reinforce this suboptimal overall process.
Several critical areas of the clinical trial lifecycle exist where AI solutions, including Machine Learning (ML) and Language Learning Models (LLMs), can alleviate these challenges. Within design and start-up, there is a massive opportunity to use components of AI during document creation and management, such as document review, trial and protocol design, consent, contracts, and study review. LLMs trained on historical documents, combined with indexing, can generate optimal index search criteria, creating a powerful tool to generate documents efficiently and find information quickly. These techniques offer massive time savings in document management and locating key data points and elements within documents. True intelligent automation or augmented intelligence, through the application of effective AI techniques, makes staff more efficient and effective, offering significant improvements in efficiency across an entire organization.
Traditional ML algorithms can be used for site feasibility, site identification, and patient recruitment. In a broader sense, when given a trial protocol, plus historical trial and site performance data, one can determine the optimal set of sites and patients for a given trial. Using ML to identify sites and optimize the identification of patients, there is a ~70% improvement in average response rates from sites willing and able to join trials versus an industry average of 30-50%. By implementing ML techniques, studies have also seen patient recruitment accelerated by ~44%.
In the area of evidence generation, such as electronic Clinical Outcome Assessment (eCOA) and electronic Patient-Reported Outcome (ePRO), monitoring for and detecting anomalies is crucial for keeping a clinical trial on track with a higher likelihood for success. Applying ML to eCOA and ePRO data in real-time as it is flowing in allows anomalies to be detected and addressed quickly and efficiently. With these techniques, clinical endpoints have a 65% reduction in error rates, drastically improving the likelihood of a trial’s success.
Through the use of AI – from ML to generative AI – clinical trials can run more effectively, efficiently, and with higher quality. The application of AI to well-known challenges in clinical trials will alleviate the issues to a great degree in the year ahead and bring life-saving drugs and therapies to society more quickly, ultimately saving lives.