The AI-Human Partnership in Clinical Trials 

Authors: Priyanka Sharma, Vice President, Software Engineering and Andrew McVeigh, Chief Architect  

Drug development is difficult, expensive, resource-intensive, and often unsuccessful. The overall success rate of trials is only 7.9%.1 Termination of pharma innovations in the R&D pipeline doubled in 2022 compared to the previous year, and an analysis of registered trials from 2000 to 2019 revealed a steady decline in completed drug trials over the decade.2,3 

While clinical trials experienced a boom during the pandemic, a 2023 report by Deloitte indicates that the effects are temporary as R&D returns are again on a decline. In fact, 2022 marked the lowest number of novel drug approvals by the FDA since 2016.4 This raises concerns for the future of public health, especially when combined with proposed funding cuts to the NIH, NCI, and CDC.5 

Given that it takes more than $2.6 billion and well over a decade to bring a new prescription drug to market, the ramifications of declining trial success may not be felt for years.6 The good news is that overall, despite the recent downturn, registered studies have increased over the years.7 However, a pressing question remains: what can we do to make clinical trials more viable and accessible? 

With the promise of faster processes at a reduced cost, the potential of artificial intelligence (AI) in clinical trials cannot be ignored. This disruptive innovation has the potential to positively impact study efficiencies that may prove transformative for the trial process and patient health outcomes in life-changing areas like oncology, infectious diseases, mental health, and many other critical areas. 

 

The AI advantage for clinical trials 

Clinical trials still have many opportunities for automation, and AI can play a role in this to help accelerate the drug discovery process and reduce costs by: 

  • Improving patient recruitment and retention: Patient data extracted from Electronic Health Records (EHRs) can be analyzed to identify eligible candidates for faster recruitment as well as predict the likelihood of patient dropouts in ongoing clinical trials. One cardiovascular study utilized AI to flag wavering participants so clinicians could provide additional education to encourage longer participation.8 AI tools can also enable participation from multiple locations with remote data collection, analysis, and monitoring. 
  • Enabling precise data analytics: AI’s ability to analyze large amounts of data can lead to faster drug discovery, quicker fails, and insights for repurposing drugs for new indications.9 Real-time trial data analysis can quickly detect trends and safety signals and inform data-driven protocol adjustments. AstraZeneca’s ongoing collaboration with BenevolentAI is one successful case that has resulted in the identification of multiple new targets in chronic kidney disease and idiopathic pulmonary fibrosis.10 
  • Optimizing trial efficiency: By reviewing large pools of historical data, AI can help sponsors design better trials, treatment protocols, and enrollment models while reducing the burden on trial coordinators.11 

 

A symbiotic partnership: AI and human expertise 

While AI’s capabilities undeniably amplify trial efficiency, the symbiotic partnership between the technology and uniquely human qualities such as critical thinking, creativity, and empathy is where clinical trials stand to benefit most. 

By taking over repetitive tasks from humans, AI can free up study teams for more high-value, specialized tasks. AI can also function as augmented intelligence, providing data that helps researchers improve treatment plans and optimize resources. The deep domain and pharmaceutical expertise of humans—understanding global regulatory environments, patient behavior, and therapeutic knowledge—is necessary to add context to interpreting AI’s data and work for optimal results. For example, researchers from the NYU School of Medicine and the NYU Center for Data Science found that combining AI with analysis from human radiologists significantly improved breast cancer detection.12 

Using AI augmentation in clinical trials can also help study teams analyze and measure unstructured data gathered from patient consent forms and clinical outcomes. These data can come in various formats, including comments on a patient’s experience. As AI still struggles with open-ended inference, human context and critical thinking are still needed for interpretation.13 AI-generated trial designs and documents, such as patient consent forms, also need human verification to ensure validity. 

We’re currently exploring these opportunities at Suvoda as we work on prototypes to analyze informed consent documents to improve patient comprehension. We’re also trialing AI as an accelerator for study design with guard rails and reviews as part of the process. In this context, AI reads the protocol, parses it, and sets trial parameters. Design consultants can then review these parameters, which are crucial for a study implementation. The building and success of this tool involves human experts in multiple disciplines, from technology to science and clinical operations. 

Regardless of AI supplementation, human connection, real-life experience, and empathy will always be essential elements in the care of patients to elicit therapeutic change. Empathy plays a crucial role in the connection between care providers and their patients. In clinical trials, participants take on considerable risk, hoping that these trials will benefit not only themselves but also provide insights to enhance health outcomes for others. Acknowledging their commitment and the time they invest is essential to honor their altruism. This human element is vital for the successful execution of clinical trials. 

 

Practical innovation for the future of successful trials 

When AI, human expertise, and technology work together, it can greatly improve the experience of those involved in clinical trials—both study participants and researchers. This has the potential to lead to more successful studies at reduced cost. It is a win-win situation, where everyone from sponsors, CROs, sites, and patients benefit from optimized workflows in clinical trials. However, much like technological innovations before, a practical approach to AI is essential to maximize its potential while minimizing the stresses of trial management and participation for study teams and patients. 

Finding the right balance will ensure that AI helps us make trials more efficient and effective while also prioritizing the safety and care of the people involved. The collaboration between AI and human insights could completely change how we approach clinical trials, bringing new levels of efficiency and innovation to clinical research. It’s an exciting time and one that we are meeting with measured optimism and a commitment to purpose-built, tested solutions that advance life-sustaining therapies for those in need. 

About the Authors 

Priyanka Sharma portraitPriyanka Sharma is Vice President, Software Engineering for Suvoda, where she leads product development, testing and site reliability initiatives. Priyanka has more than 20 years of experience driving software innovation in highly regulated industries, with a focus on establishing best practices to ensure global scalability and compliance. Priyanka can be reached via LinkedIn.  

Andrew McVeigh portraitAndrew McVeigh is Chief Architect for Suvoda, where he leads the design and build of Suvoda’s next generation platform and ecosystem to manage the patient journey for the most complex clinical trials. He is a technology industry veteran, having held leadership roles at Amazon, RiotGames, Hulu and LiveRamp. Contact Andrew through LinkedIn


 

REFERENCES 

  1. Kim, E., Yang, J., Park, S. et al. “Factors Affecting Success of New Drug Clinical Trials.” Ther Innov Regul Sci 57, (2023): 737–750. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10173933/
  2. “Seize the Digital Momentum Measuring the Return from Pharmaceutical Innovation 2022.” Deloitte Center for Health Solutions, 2023. 
  3. Gresham G, Meinert JL, Gresham AG, Meinert CL. “Assessment of Trends in the Design, Accrual, and Completion of Trials Registered in ClinicalTrials.gov by Sponsor Type, 2000-2019.” JAMA Netw Open. 2020;3(8):e2014682. https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2769830 
  4. “New Drug Therapy Approvals 2022.” U.S Food & Drug Administration, January 10, 2023. https://www.fda.gov/drugs/new-drugs-fda-cders-new-molecular-entities-and-new-therapeutic-biological-products/new-drug-therapy-approvals-2022 
  5. “House Subcommittee Considers FY24 Appropriations Bill that Cuts Cancer Research and Prevention.” American Cancer Society Cancer Action Network, July 14, 2023. https://www.fightcancer.org/releases/house-subcommittee-considers-fy24-appropriations-bill-cuts-cancer-research-and-prevention 
  6. “The Pursuit of Excellence in New-drug Development.” McKinsey & Co, November 1, 2019. https://www.mckinsey.com/industries/life-sciences/our-insights/the-pursuit-of-excellence-in-new-drug-development#/  
  7. “Number of Registered Studies Over Time.” ClinicalTrials.Gov. U.S National Library of Medicine, Accessed September 8, 2023. Available at: https://classic.clinicaltrials.gov/ct2/resources/trends#RegisteredStudiesOverTime  
  8. Askin, S., Burkhalter, D., Calado, G. et al. “Artificial Intelligence Applied to clinical trials: opportunities and challenges.” Health Technol. 13, (2023): 203–213. https://rdcu.be/dlDQN 
  9. “Intelligent Drug Discovery Powered by AI.” Deloitte Insights, November 10, 2019. https://www2.deloitte.com/us/en/insights/industry/life-sciences/artificial-intelligence-biopharma-intelligent-drug-discovery.html/#endnote-8 
  10. “AstraZeneca takes home 2 more computer-generated drug targets from BenevolentAI.” FierceBiotech, October 6, 2022. https://www.fiercebiotech.com/medtech/astrazeneca-takes-home-two-more-computer-generated-drug-targets-benevolentai11 
  11. Vora LK, Gholap AD, Jetha K, et al. “Artificial Intelligence in Pharmaceutical Technology and Drug Delivery Design.” Pharmaceutics 15, no.7 (2023): 1916. https://www.mdpi.com/1999-4923/15/7/1916 
  12. N. Wu et al., “Deep Neural Networks Improve Radiologists’ Performance in Breast Cancer Screening” IEEE Transactions on Medical Imaging 39, no. 4 (2020): 1184-1194. https://ieeexplore.ieee.org/document/8861376 
  13. Marcus G. “Deep Learning: A Critical Appraisal.” 2018. arXiv:1801.00631 [cs, stat]. https://arxiv.org/ftp/arxiv/papers/1801/1801.00631.pdf