While 2024 might not eradicate the lack of representation in clinical trials, thanks to the integration of AI, it will be a pivotal year where significant strides are made. Healthcare leaders have an unprecedented opportunity to harness the potential of AI to address healthcare disparities, particularly within the realm of clinical trials. Here, we explore five ways AI is poised to transform clinical trials.
- Identify underrepresented populations
With the Rise of AI, What IP Disputes in Healthcare Are Likely to Emerge?
Munck Wilson Mandala Partner Greg Howison shared his perspective on some of the legal ramifications around AI, IP, connected devices and the data they generate, in response to emailed questions.
Clinical research often fails to reflect diverse populations, leading to an incomplete understanding of the effectiveness of treatments. A U.S. study of over 3,000 patients enrolled in cancer trials revealed that Black and Hispanic patients had lower Phase I enrollment. The underrepresentation of certain groups in clinical trials poses the risk of overlooking differences in drug metabolism, side effect profiles, and outcomes. This omission can lead to harmful responses to therapies and an incomplete understanding of treatment effectiveness.
AI can play a crucial role in identifying underrepresented populations in clinical trials by quickly analyzing vast amounts of existing healthcare data. By leveraging machine learning (ML) and AI, researchers can gain insights into patient demographics, genetic profiles, and other healthcare data to understand and address the underrepresentation of specific populations. This information can guide researchers and trial organizers to actively target and engage specific demographics that may have historically been overlooked or underrepresented.
- Optimize trial design & site selection
Selecting the right site, breaking down participation barriers, projecting accurate enrollment numbers, and maintaining consistent communications between principal investigators (PIs) and participants are all critical to a trial’s success. AI optimizes all of these processes to ensure that trial protocols, eligibility criteria, and recruitment efforts are more inclusive from the outset.
A Personalized Approach to Medication Nonadherence
At the Abarca Forward conference earlier this year, George Van Antwerp, managing director at Deloitte, discussed how social determinants of health and a personalized member experience can improve medication adherence and health outcomes.
By analyzing historical trial data and taking into account demographic factors, AI can help researchers identify ideal trial sites and PIs/clinical research organizations (CROs). AI can also help pinpoint community research sites that hold trusted relationships with patients who are often overlooked during the trial process.
Furthermore, AI can be leveraged to identify the potential barriers to participation for diverse patients, and AI-powered devices can help close the gaps. For example, according to a Deloitte Insights report, the primary obstacle to diverse clinical trial participation is access. AI-powered wearable devices serve as a transformative solution by minimizing the need for participants to physically travel to trial sites. This enhances accessibility for individuals eager to engage in these trials, helping to improve recruitment and participation of diverse patient populations.
- Turbocharge patient engagement & recruitment strategies
Patient recruitment is often a major bottleneck in clinical trials, taking significant time and resources. Indeed, up to 29% of Phase III trials fail due to poor recruitment strategies. AI can speed up these processes, predicting patient availability based on historical data and detecting and mitigating biases in trial recruitment processes to make efforts more successful.
AI-powered algorithms can quickly analyze a broad range of factors beyond just demographic and health data—including socioeconomic status, cultural background, and geographic location—to identify ideal clinical trial participants. These insights enhance decision-making and enable researchers to design more inclusive recruitment strategies based on diverse factors.
Leading pharmaceutical companies like Amgen, Bayer, and Novartis are at the forefront of leveraging AI. They are actively training AI systems to analyze vast datasets, including billions of public health records, prescription data, and medical insurance claims. This approach not only streamlines the identification of potential trial patients but, in some instances, has reduced enrollment time by half.
Furthermore, the power of AI can help deliver transformative, person-centered care. GenAI-based insights help clinicians develop tailored recommendations on the “next best action”— the best way to engage various patient populations in a culturally relevant manner.
- Enable real-time monitoring and adaptive trials
AI enables real-time monitoring of trial participants via wearable devices and sensors, allowing for immediate identification of any disparities or biases that may emerge during the course of the trial.
AI tools can also be used to monitor site performance once the trial has started to detect adverse events and predict outcomes, allowing researchers to identify potential issues or trends early in the process. One study found that ML prediction models reduced cancer mortality by 15–25% across several clinical trials, and also found evidence of ML algorithms supporting early detection and prognosis of disease, thus improving overall trial success.
This synchronous feedback loop enhances trial efficiency and efficacy by allowing for adaptive trial design where protocols can be adjusted to address issues, ensure equity in participant representation, prioritize patient safety, and improve overall success in developing new treatments.
- Tackle biases in data collection
In the context of healthcare and clinical trial data, mitigating bias is crucial to ensure the effectiveness, fairness, and safety of medical treatments. AI holds the potential to eliminate long-standing biases in healthcare data, particularly in Electronic Medical Records (EMR) and Electronic Health Records (EHR).
When implemented and trained properly, AI systems will avoid perpetuating biases and help improve data collection methodologies to ensure diverse populations are accurately represented. One of the key challenges is the lack of diversity in clinical datasets, which can lead to biased AI algorithms. If the training data is misrepresentative of the population, AI is prone to reinforcing bias, potentially leading to undesired outcomes or misdiagnoses. To address this, AI can synthesize underrepresented data and detect biases in the data collection and preparation stages, thereby creating technology that is fairer and more accurate. Furthermore, by involving clinicians in data science teams, a broader perspective is attained and bias can be prevented at various stages of algorithm development and monitoring.
The (slightly bumpy) road to success
The integration of AI technologies holds promise for enhancing outreach efforts, streamlining recruitment processes, and addressing long-standing barriers and biases that hinder diversity and inclusion in clinical trials. Still, there are roadblocks to its effective implementation, including resistance to change or distrust, security concerns, high costs to develop custom systems, and proper usage guidelines and staff training.
The biggest challenge delaying widespread adoption and success is improving the breadth, quality, diversity, and accessibility of the underlying data, on which these AI systems are trained. Without addressing this head on, we will continue to see biases perpetuated and hallucinations that contain false or misleading information.
There are a number of promising federal efforts underway to help guide us, such as the FDA’s guidance around diversity action plans for clinical trials, the President’s executive order on the use of AI, the FDA’s plans to establish a Digital Health Advisory Committee, and the EU’s AI Act. It will be crucial for leaders to align AI use with these emerging regulations. By taking the right steps, it is possible to create AI systems that are beneficial for all and will positively transform clinical trial processes, ultimately contributing to the reduction of healthcare disparities.
Photo: Sylverarts, Getty Images
Ryan Brown is Regional Vice President, Sales-Trial Landscape at H1. Trial Landscape is H1’s exhaustive clinical trial intelligence repository, incorporating data from public and proprietary sources including over 10 million healthcare providers (HCPs) and over 420,000 clinical trials. It is the first solution of its kind to fully integrate diversity and inclusion insights at the site, HCP, patient, and now, the indication levels – accelerating site and PI research, validation, prioritization, diversity, and selection. Ryan is passionate about improving health care equity, access and outcomes in clinical research for the patients we serve through the vehicle of diversity.
