A Perspective on Possible Applications of Artificial Intelligence to the Clinical Trial Workforce

Clinical trials provide the scientific foundation for justifying the safety and efficacy of drugs, biologics, and devices—but are laborious, expensive, and risky. Only 10% of drugs entering clinical trials receive U.S. Food and Drug Administration (FDA) approval, while common reasons for trial failure include poor patient selection, recruitment challenges, and complex study designs.{1} The pharmaceutical industry has begun leveraging artificial intelligence (AI) to streamline numerous aspects of drug development from identification of novel drug targets to clinical trial design.{2} Moreover, AI is being leveraged to predict clinical parameters ranging from disease onset to mortality, drug-target connections, and drug repositioning suggestions.

In response, the FDA is devising an ongoing regulatory framework that will consider feedback from various stakeholders within drug development to spur innovation, while promoting patient safety.{3} The clinical research professionals working behind the scenes play a vital role in the drug development process, yet face administrative burden, burnout, and high turnover rates that all adversely impact clinical trial quality and the safety of new therapies.{4}  For instance, a little more than half of the U.S. clinical trial workforce reported burnout since the 2020 COVID-19 pandemic.{4}

AI could be leveraged to reduce the administrative burden associated with laborious, site-specific documents to provide research teams more time to conduct high-priority tasks such as identifying, screening, and enrolling patients. For instance, AI could analyze a clinical study protocol in PDF format and generate a series of documents, forms, and templates to aid with study execution based on insights and similarities drawn from a database of similar protocols. Such documents could include study-specific checklists, guides, and notes to assist the clinical team in conducting a study more efficiently.

Furthermore, machine learning tools could build protocol-specific electronic case report forms for data capture using existing protocols and databases to establish a new electronic data capture (EDC) database. This could generate EDC builds faster than traditional teams that could then be reviewed by data management teams for discrepancies. Clinical study report (CSR) generation can also benefit from AI-assisted product information management by producing initial draft shells of a clinical study report based on each individual protocol’s specifications. Automating this task alone will garner significant time savings at end of study by expediting submissions to, and ultimately approvals from, drug authorities.

However, implementation of AI models to clinical trials presents risks that must be considered since sensitive patient data and sponsor proprietary information are involved. Potential risks include incorrect data computation resulting in protocol deviations; data breaches that may disclose confidential information; and various legal and ethical implications pertaining to data privacy and compliance with country-specific regulations—such as General Data Protection Regulation in the European Union and the Health Insurance Portability and Accountability Act in the United States.

Addressing these risks must include a multi-disciplinary approach that involves legal experts and experienced AI developers to create a robust core that will protect patients while encouraging new innovations in clinical trials. With strong data encryption and access controls/monitoring, an AI model would need to ensure that data are encrypted when being transferred between systems. Including an additional security measure can safeguard clinical trial data. Implementing access to the AI model should be rigorous to ensure that only authorized clinical trial team members can access the system. Continuous monitoring of access logs can assist in detecting and preventing unauthorized access to the AI model.

Conclusion

AI will profoundly impact numerous industries, and clinical research is no exception to this technological shift. Industry leaders have focused on utilizing AI to improve clinical trial design, patient recruitment, and data quality. For instance, a recent study focused on the implementation of an AI tool called Trial Pathfinder, using real-world data (RWD) to simulate data for non-small cell lung cancer. This then allowed the AI tool to learn both clinical protocols and patient RWD.{5}

While these approaches are undoubtedly important, another area requiring attention involves actual trial operations. We believe that technology should also be used to reduce the burden on study teams, whether at a clinical trial site, contract research organization, or study sponsor. Using an AI system to generate study- and site-level documents, in addition to building an electronic database to automatically create CSRs, would contribute to delivering lifesaving therapies to patient populations faster.

Submitted as independent authors by Justin Scott Brathwaite, PMP, CCRP; Lindsey Haroun; Daniel Goldstein, MSN, RN, CCRP; Erin Dowgiallo; Karah Hogue; John Seeley, MBA

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding Sources

None

References

  1. Van Norman GA. 2016. Drugs, devices, and the FDA: Part 1. JACC Basic to Transl Sci 1(3):170–9. https://doi.org/10.1016/j.jacbts.2016.03.002
  2. Bhatt A. 2021. Artificial intelligence in managing clinical trial design and conduct: man and machine still on the learning curve? Perspect Clin Res 12(1):1. https://doi.org/10.4103/picr.picr_312_20
  3. U.S. Food and Drug Administration. Center for Drug Evaluation and Research. 2023. Artificial intelligence and machine learning (AI/ML) for drug development. https://www.fda.gov/science-research/science-and-research-special-topics/artificial-intelligence-and-machine-learning-aiml-drug-development
  4. Causey M. 2021. Burnout threatens to undermine clinical trial workforce productivity. ACRP Blog. https://acrpnet.org/2021/09/burnout-threatens-to-undermine-clinical-trial-workforce-productivity/
  5. Zhang B, Zhang L, Chen J, Jin Z, Liu S, Zhang S. 2023. Harnessing artificial intelligence to improve clinical trial design. Commun Med 3(1). https://doi.org/10.1038/s43856-023-00425-3