Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Clinical Trials with AI: Chasing efficiency and inclusive participation

Innovations using AI in the field of clinical trials are here to stay. But, with great power comes great responsibilities.

Woman looking through a microscope

[Images: Adobe Stock / metamorworks]

Profile photo of Avik Pal

Heightened awareness, discussions, and experimentation of the use of artificial intelligence (AI) in healthcare and life sciences are progressing at a breakneck pace.

While rules and regulations around the use of AI are still a work in progress, the potential of AI to speed up medical research for better patient outcomes is already evident. Recently a USFDA guidance highlighted the benefits of AI in clinical research and the use of technologies to improve diversity and inclusion in clinical research that impact the generalizability of study results. The time has come for us to take a deeper dive to understand how AI can make clinical trials (CTs) more efficient and how AI’s transformative power can increase the diversity and inclusion of underserved populations as study participants. 

AI FOR EFFICIENT CTs 

There is an enormous amount of clinical, molecular, and imaging data collected daily in hospitals and clinics that can generate rich insights critical for the advancement of medical research. However, generating actionable insights from these largely uncategorized and complex data is a formidable challenge. Comprehensive AI models trained with appropriate data sets can yield automated meaningful insights that can expedite clinical trials and improve patient outcomes.

Clinical trial teams may use predictive analysis to identify new molecules that have a higher probability of successfully meeting primary outcome(s) based on study design and eligibility criteria of study participants.

Additionally, AI tools can also help match patients to eligibility requirements using demographic, genomic, and geographical data from patient profiles. 

Decentralized clinical trial (DCT) technologies such as AI-powered smart wearable devices can monitor trial participants from the comfort of their homes. This ensures better adherence to protocol and timely reporting of adverse events. AI models can be trained to detect, for example, an impending cardiac arrest prior to the development of symptoms, based on biometrics of the patient under surveillance.

AI FOR MORE INCLUSIVE PARTICIPATION IN CTs

Underserved populations have historically been limited in their participation in clinical trials, and therefore may not be able to leverage the benefit from the therapies that are being studied. A high burden of disease among the underserved is often indicated in many demographic analyses of the patient population. For example, of the 4,922 patients in the U.S. who participated in trials that led to the approvals of 18 new drugs in 2022, 73% of patients were White, 5% were African American, 14% were Asian, 6% were Hispanic.

However, data shows that in the U.S., particularly in cancer patients, there are racial and health disparities: African American men and women have a 111% and 39% higher risk of dying from prostate cancer and breast cancer, respectively, compared to their white counterparts; Hispanic children and adolescents are 20% and 38% more likely to develop leukemia than non-Hispanic white children and adolescents, respectively; and Asia/Pacific Islander adults are twice as likely to die from stomach cancer as white adults.

Digital innovators in the field of clinical research are currently facing the challenge of developing solutions that can enable wider patient participation for more inclusive outcomes that can address health inequities. Advanced AI-enabled clinical research technologies can help widen the net for more diverse patient recruitment. Vast data from electronic medical records systems at hospital and clinical sites can serve as rich resources for training AI and machine learning tools that can eventually search and match eligible patients to studies faster.

CHALLENGES OF IMPLEMENTING AI IN CTs

A significant challenge for AI developers is to prevent any perpetuation of bias in the data sets for AI training. Since historically, the demography of CT participants is skewed towards the white population, more representative data sets are needed for training AI tools to appropriately match the patient population of diverse ethnicities with the eligibility criteria of CTs. Awareness is needed amongst CT teams to diversify their trial participants to reap the benefits of AI models in the near future.

Innovations using AI in the field of clinical trials are here to stay. But, with great power comes great responsibilities. Hence, there is a need for robust safeguards to protect the security of personal data. While the mission to have more efficient clinical trials and more inclusive participation is laudable, the advocates, the author included, are fully aware of the risks of unregulated use of AI and are working hard to ensure AI innovation follows regulations and complements human expertise. In short, we need to ensure the responsible and ethical use of AI, irrespective of whether regulation is in place or still in progress.


Avik Pal Founder & CEO, CliniOps, Inc.