The history of diversity in clinical trial populations tells a stark story about medical research in America. Despite decades of regulatory pressure, clinical trials too often exclude some of the very populations they're meant to serve. Women, members of racial and ethnic minority groups, rural patients, and older adults remain significantly underrepresented in studies that determine whether new drugs and devices work‌ — ‌and for whom.

Under traditional methods, when cancer treatments are studied mainly in urban academic medical centers, the needs of rural patients might never be reflected in‌ life-saving therapies. And when a cardiovascular drug is tested primarily on white men in their 50s, doctors don’t know how it might work for a 70-year-old Latino woman with diabetes.

It’s not only a matter of selective recruitment. "Some groups are just far less likely to participate in clinical trials,” said Rachel Richesson, a clinical professor of learning health sciences at the University of Michigan Medical School. “They might be less trusting of the system. They might have fewer resources. They often can be more time limited. They might have transportation issues. And our drug development processes haven’t yet figured out how to clear those obstacles."

To overcome those longstanding challenges, pharmaceutical companies and contract research organizations (CROs) — which conduct clinical trials for companies that develop medications and medical devices‌ — ‌are turning to ‌new technologies configured for clinical trials. The trend is nascent but decidedly promising, for patients and pharmaceutical companies alike.

Solutions are now being fine-tuned for clinical trials, to help overcome this fundamental public-health problem. The technology can integrate information from a variety of disparate sources, such as patient registries, claims data, electronic medical records, population health platforms, and device-specific data feeds, to name just a few. And once that information is pulled together, AI tools can enable the organizers of clinical trials to make strategic use of what might otherwise be a welter of bewildering data.

A more inclusive setup would be an AI-powered integrated clinical trial platform connecting the healthcare ecosystem so that "as a pharma company I come in, design my clinical trial protocol, I design my data-gathering forms and execute my trials with holistic visibility,’’ said Siddhartha Bhattacharya, who specializes in healthcare operations, AI, design and product management at the management consultancy PwC. “It would be a huge benefit,”  Bhattacharya said, “because it can give pharma companies visibility into the overall process and enable healthcare providers and patients' flexibility to participate in a trial."

Historical barriers to clinical trial diversity

For decades, researchers have defaulted to recruiting from easily accessible populations ‌ — ‌ typically white, male, urban patients who could regularly visit major medical centers. In the past, there was a bias against including women on grounds that menstrual cycles could complicate the findings. And abusive research like the federal government’s 40-year “Untreated Syphilis Study at Tuskegee” had made many Black Americans wary of medical studies.

Economic pressures have compounded the problem. Getting new drugs from the laboratory to market takes between 10 and 15 years, and costs more than $2.5 billion on average, creating incentives to prioritize efficiency over inclusivity.

"If a drug is second to market in its target indication, it can mean a significant dropoff in revenue potential for the pharma company, " explained Sharmin Nasrullah, General Manager of Life Sciences Clinical Development at Salesforce. "Being first in this race maximizes market share and provides a longer period of exclusivity."

Starting up each clinical site also costs millions of dollars, pushing companies toward familiar, high-performing locations rather than venturing into new communities. The result is a self-perpetuating cycle: Trials too often fail to represent real-world patient populations, limiting evidence for how treatments worked across different groups.

Another challenge: Reaching a prospective patient pool representative of the broader population.

“One of the hardest parts of recruiting patients is raising awareness among historically underrepresented patient groups about the existence and value of clinical trials,’’ said Ali Ahmed, Salesforce Global Head of Industry Innovation for Pharmaceuticals and Therapeutics. That’s why recruitment for clinical trials, he said, should use the same digital techniques that consumer-products companies now rely on to reach the broadest audience.

“We can create opportunities to engage with a whole new demographic of patients,” Ahmed said, “whether it's through social media or mobile or digital capabilities that, traditionally, pharmaceutical companies just didn’t even think about.”

Overcoming disconnected data

Another major obstacle to clinical trial diversity is data fragmentation. Clinical trial operations today rely on disconnected systems, each holding pieces of the recruitment puzzle but unable to communicate with others to piece it all together.

“The clinical trial and marketing platforms are often disconnected from each other,” said Magon Mair, Director of Solution Engineering for Wilco Source, a company that helps pharmaceutical companies implement clinical trial and other life sciences industry solutions using Salesforce technologies. The phone systems are separate. The websites are separate. Their email isn’t connected to their CRM.”

”Even when potential clinical trial participants do express interest,” Mair said, responses typically “take one to three days with some CROs right now” as a result of those disjointed data systems. “As a patient,” she said, “my expectation is to get an email back quickly while I'm excited and interested."

Not only do these disconnected systems cause delays, Mair said, but they also prevent recruiters from answering basic questions about potential recruits: Do they prefer texts or phone calls? What’s their preferred language? What other trials might they qualify for? The data exists somewhere. But when scattered across platforms, it remains practically useless.

The AI advantage

Unified data also improves AI outputs. For example, by drawing up-to-date information from a variety of an organization’s systems, an AI agent can more effectively synthesize information to match participants with the right trials or identify the best sites for a given study.

Mair has customized multiple Salesforce Life Sciences Cloud to help CROs do just that. “They can take all those different forms of data wherever it's coming from‌ — ‌be it a form, a phone call, an email, even a piece of paper‌ — ‌and consume that information and make it meaningful," she said.

The impact can be immediate. Mair demonstrated how CROs can reduce the participant response time to 15 minutes from three days. For patients dealing with serious illnesses, that responsiveness can mean the difference between enrollment and a missed opportunity.

Likewise, if a patient from an underrepresented ethnic group “screens out” of one trial‌ — ‌say, a diabetes study that excludes smokers‌ — ‌AI tools can immediately identify other research opportunities, like a hypertension study that needs smokers from that patient’s ethnic group.

With the latest patient data, AI systems can also analyze recruitment patterns in real time, proactively flagging demographic imbalances to trial operators before they become problems, Mair said. This capability, in particular, is embedded in Salesforce’s Agentforce platform, which helps healthcare and life sciences companies build and deploy AI agents that automate tasks.

 "The AI can pull information that says, ‘We’ve got too many people on this trial who are 40 to 50 years old,’” she said. “Go find people who are 20 to 40 who qualify, so we can balance this out.'"

Meeting patients where they are

The most promising applications combine technological efficiency with a deeper understanding of human behavior and community needs — right down to local logistical challenges. Clinical trials conducted during the COVID-19 pandemic demonstrated that the studies could extend beyond traditional medical centers to include telehealth visits, home-based care, and partnerships with local healthcare providers.

This decentralized approach has created new opportunities for inclusion, but also additional data integration challenges: Information now flows from wearable devices, home visits, local clinics, and community health centers.

Platforms like Life Sciences Cloud and Agentforce can not only absorb and synthesize data from these previously inaccessible sources. They can use it to autonomously gather previous generations of clinical trial recruiters could only dream of —  like identifying patients at risk of dropping out and identifying the obstacles they are facing. Then AI agents can proactively solve those problems on the patient’s behalf.

Take, for instance, a patient who lives in a neighborhood where buses typically run late on weekday afternoons. Life Sciences Cloud’s recruitment technology can identify that as a risk factor for patients in that neighborhood who rely on public transportation to get to their appointments on time, and arrange for an Uber or Lyft to pick them up instead (at no expense to the patient).

The patient gets notified of the arrangement via email or text — again, depending on their preference — and their risk of dropping out plummets. “This is the kind of personalized patient engagement that’s needed to meet patients where they are,” Nasrullah said.

Richesson, of the University of Michigan, sees potential for sophisticated community-based applications of the technology as well. "You could potentially look across the system and see how people reached out to participants, like how many contacts it took a research coordinator to reach someone,” she said. “You might have to make 15 calls to recruit someone from a particular geographic area or patient population."

By analyzing these patterns and combining them with community data, the technology could help researchers develop community-engagement strategies for different populations, she added. “Connection-building and trust are paramount,” Richesson said. “This is still fundamentally a human problem. We're asking people to give up their time and take a risk for us."

Trending in the right direction

Early results suggest the industry’s adoption of data-and AI-driven technologies is paying off. Mair, for instance, noted that schedule-to-show rates‌ — ‌how often patients keep their appointments‌ — ‌can improve by 25 to 50% when patients receive automated reminders. And she cited how CROs are leveraging new technologies to double the number of study sites the organization could manage without increasing staff.

Such results could foretell even greater investments among CROs and pharmaceutical companies in the future. For an industry that has long struggled with competing demands of scientific rigor, economic pressure, and social responsibility, integrated, AI-powered technology platforms offer a path toward clinical trials that are both more efficient and more inclusive.

Rather than treating diversity as a regulatory checkbox, these tools make it possible to design trials that are more fully representative from the outset. That’s the best way to ensure the next generation of medical treatments will work for all patients, not just the ones who are the most convenient for researchers to recruit. 

"Because people are inherently diverse in all kinds of different ways - across demographics, wealth, race, education — to crack the code on recruitment, we actually have to offer multiple methods of recruitment and meet patients where they are," Nasrullah said. "Patients are diverse. Therefore, recruitment methods must be diverse."

Go deeper:

• Learn more about Agentforce for Life Sciences Cloud
• Watch a Life Sciences Cloud demonstration
• Read about the future of clinical trials
• Read about Salesforce’s robust partner network and clinical customers