Find care now
If you are experiencing a medical emergency, please call 911 or seek care at an emergency room.
Clinical trials are an essential way we create new knowledge about how to help our patients heal. Yet, despite our efforts to expedite this work, clinical trials remain inefficient, expensive, slow, and plagued by inequities. We believe artificial intelligence can be part of the solution to help make these trials more efficient, more accessible, and more equitable. In fact, I’m optimistic that AI will bring about the biggest changes in how clinical trials are conducted since the dawn of the internet.
A clinical trial tests new medical treatments like drugs and devices to learn whether they are safe and effective for people to use and whether they improve outcomes over conventional medical practice. In January, the National Library of Medicine recorded more than 20,400 trials recruiting participants in the U.S. The largest percentage of these, about 15%, were related to cancer, with mental health, endocrinology, and cardiovascular trials each accounting for an additional 5-6%.
At MedStar Health Research Institute, investigators have been among the first to leverage AI in research to process huge amounts of data, root out unconscious bias, and expand access—especially when it comes to issues that matter to our community, like maternal health and health equity. We also conduct hundreds of clinical trials and are among those who recognize that the traditional approach to trials is long, complex, and inefficient. That is why we became early member of the Clinical Trial Transformation Initiative and continue to seek ways to make clinical trials for efficient and equitable for all in our community.
Because of our early adoption of AI in research and our decades of experience in clinical trials, we believe we can improve clinical research for everyone through AI applications. Well-regulated, transparent, strategically applied artificial intelligence can provide enormous benefit to our patients by improving the clinical trial process.
AI to improve clinical trial efficiency.
The National Institutes of Health (NIH) is the largest public funder of medical studies in the world. In 2023 alone, the NIH funded $18.9 billion in clinical research. Pharmaceutical and medical device companies also invest billions to fund clinical trials. So much funding is necessary, in part, because making clinical trials safe and effective is very complex—and currently quite inefficient.
There are several reasons clinical trials have become complex. To ensure trials are focused on benefitting specific diseases and/or specific patient populations, detailed ‘inclusion’ and ‘exclusion’ criteria are developed. This means only a small percentage of patients with an illness are eligible to be enrolled in a research study. For example, a study for a new chemotherapy for breast cancer may exclude patients with high blood pressure, diabetes, or allergies and may also set limits on the eligible patients’ age or type of breast cancer. Complicated criteria can make it challenging to enroll enough patients into a study to make it worthwhile.
Another challenge of clinical trials is the multitude of regulations. There are several regulations and multiple agencies that provide oversight of trials which can result in a tremendous amount of administrative burden. Furthermore, an audit trail with source documentation is required and studies are often reviewed by study ‘monitors’ from the sponsor. The long paper trail is an important check on research but also adds significant complexity and time to our work.
The end result of these and other inefficiencies is that it takes twice the money and twice the time it should to study and approve new drugs and devices to help our patients. That’s a big part of why it can cost an average of nearly $1 billion to develop a new drug.
We believe AI can help.
AI can scan enormous volumes of data in a short time, revealing insights that can provide a more effective means of helping patients identify trials that are right for them. It can also work in reverse, helping investigators identify which patients might fit their criteria and benefit from their work.
The design of trials can also benefit from AI assist. Now, researchers design their trial methods based largely on their expert experience with, at most, a few dozen of trials. AI can look across thousands of studies to recommend the most efficient design and understand which patients are the best fit. The result is smarter trials open to more patients.
Related reading: Publication to Practice: How Research Drives Clinical Action to Improve Patient Care
AI assistance can make trials more accessible.
AI can help make the benefits of clinical trial participation, like early access to the latest treatments, available to more patients by helping link participants with studies. It can also help us understand how to help people stay in trials by improving communication. Studies have found that 15% - 40% of trial participants drop out of the trial before it’s complete. Without participants, a study can’t accomplish its goals, but there are many reasons why they might choose to withdraw. AI can help with at least one of these: education and engagement about the trial’s implications for their health and medical knowledge at large.
At MedStar Health Research Institute, we have found AI powered chatbots can be a very effective means to educate patients with an eye toward cultural competency and responsiveness. We’re studying how chatbots can help our patients learn about their condition and care in several settings, from maternal health to heart failure, cancer treatment, and more.
Similarly, a study this spring found that AI-generated communications were more readable and empathetic than letters from physicians. Generative AI algorithms can produce integrated communications that match tone and consider cultural sensitivities, making programs like chatbots a powerful tool. Because they’re able to respond to patients in real time, AI-powered chatbots can be exceptionally good communicators. This technology can be a game changer in patient communication, satisfaction, and retention, making the benefits of trials more available to more people and keeping more trials active.
Related reading: Health Systems and Researchers Can, and Must, Work to Advance Community Health and Equitable Access.
Rooting out unconscious bias with AI’s help.
Every conversation about AI includes the importance of ensuring the technology is safe, reliable, and trustworthy. That goes double for AI’s involvement in clinical trials. After all, it’s one thing to have AI pick out the next song you should hear. Helping you know when to go to the doctor or participate in a clinical trial is much more serious.
The way to make sure AI algorithms are safe and reliable is by testing them rigorously in a real-world environment. MedStar Health is just such an environment. Because we are a large healthcare system, we serve many types of patients in many settings with a very diverse pool of caregivers. Our diversity mirrors that of the nation, so AI that works here will work elsewhere.
MedStar Health is well-positioned to provide assurance testing for AI products. Our National Center for Human Factors in Healthcare is experienced in usability testing, and our Center for Health Equity Research looks across races, ethnicities, and gender identities to understand how healthcare studies impact us all. We are applying our full expertise in these areas when we test and apply AI to research studies.
When it comes to equity, we work diligently to understand how implicit bias impacts care. One 2015 study found that up to 70% of physicians in the United States have at least some unconscious bias that impacts care of Black and Latinx patients. This can perpetuate systemic inequalities, worsening care for patients.
Generative AI has already shown an impressive ability to communicate with cultural nuance. That means it can also be used to detect hints of implicit bias in provider notes and the ordering and interpreting of tests. When we can detect bias, we have an opportunity to intervene to improve outcomes, enhance care, and build more trust with our communities, including clinical trials that are more inclusive and devoid of bias.
Related reading: How Research Can Drive Improved Healthcare Safety and Equity When Using AI.
AI will change the face of clinical trials.
When I was a young cardiologist, my bookshelves were crammed with the latest textbooks so I could always access the most up-to-date knowledge to benefit my patients. Then the internet came along. I found myself becoming an author in a digital textbook that gets updated every few months. So, while I still have a lot of books, the truth is I turn to a computer or smartphone when I want to get the latest in medical advances. AI can have a similarly groundbreaking impact. It is an insightful copilot that is ready to help us provide better care and conduct better research.
As we progress, AI is going to point out things we’re missing and things we don’t realize we’re doing. It will give us suggestions we can use to provide the best patient care. It’s going to make trials more efficient, open, and trustworthy, and it’s going to support our work of discovering the next generation of drugs and devices that improve everyone’s health.
