Tips for Optimizing Patient Recruitment in Clinical Trials

Patient Matching for Clinical Trials Simplified with LLMs and FHIR Securing eligible participants for clinical trials is a complex and costly process. As much as 32% of a trial's total budget can be spent on patient recruitment alone, a significant portion of which is spent on manual review to determine patient eligibility. The primary challenge is clinical protocols, where inclusion and exclusion criteria are typically written as unstructured, free-text documents. Matching these detailed, often nuanced criteria against a patient's structured and unstructured data in the EHR requires manual reviews. Many in the industry are exploring using large language models (LLMs) to address these challenges. In a recent video below, we shared how we are exploring combining the power of LLMs with the structured data in FHIR format. We demonstrated two key scenarios for this combined LLM-FHIR approach: • Single-Patient Eligibility: This scenario focuses on an individual patient's chart. Once a patient has been identified as a potential candidate for a trial (e.g., through a high-level condition-based query), the system can perform a detailed analysis. The LLM extracts the inclusion and exclusion criteria from the unstructured protocol. It then uses a combination of structured FHIR queries and LLM-based matching with patients' data, including structured data points (like lab results) and unstructured clinical notes. • Multi-Patient Eligibility (Bulk Matching): In the second approach, we leverage FHIR Groups to screen a cohort of patients against a trial's criteria simultaneously. The system performs the same LLM-driven criteria extraction and FHIR querying process but on a larger scale, providing a summary report of eligible and ineligible patients within the group. Combining LLMs and FHIR offers a powerful way to streamline clinical trial recruitment. This approach can also be applied to other use cases, such as matching many trials to many patients or finding matching trial(s) for a patient. With patient access to their data now becoming simpler, this also opens up opportunities for patient-initiated self-matching for trials. Finally, this can also be used to find patients for internal research studies. We are exploring these scenarios at Darena Health. If you are working on similar challenges or are looking to implement a solution like this, we would love to compare notes and see if there are collaboration opportunities.

After nearly a decade in the trenches of Alzheimer’s trials, certain patterns become hard to ignore. Whether it’s a small biotech or a global sponsor, we keep seeing the same preventable pitfalls stall recruitment. We’ve consolidated our insights into a new whitepaper: Common Mistakes in Recruiting for Alzheimer's Disease Clinical Trials: A Strategic Framework for Optimization. The six mistakes inside are preventable, and we keep seeing them, across sponsors, across programs, across continents. A few things inside that I think the field needs to hear: → The MMSE has a ceiling effect that makes it nearly useless for preclinical populations. We have better tools now. → Tiered screening: digital cognitive assessments first, then blood-based biomarkers, then PET can cut recruitment costs and patient burden. → Community-based recruitment isn't just a nice-to-have for diversity. It's operationally necessary to find asymptomatic participants who will never walk into a memory clinic as the field moves into earlier stages of the disease. → Real-time data beats scheduled reports. Recruitment managed dynamically outperforms "set it and forget it" every time. The whitepaper includes perspectives from sponsors, site coordinators, and data from Eli Lilly and Company TRAILBLAZER-ALZ 3 program. Thank you to Alan Kott (Signant Health), Lennert Steukers (Johnson & Johnson Innovative Medicine), Ralph Lee (Irvine Clinical Research) for their valuable contributions. If you're working on AD trials, I'd encourage you to read it. Link in comments. #AlzheimersResearch #ClinicalTrials #DrugDevelopment #DigitalBiomarkers #ClinicalRecruitment

Enhancing patient participation in clinical trials isn't about putting them at the centre—it's about creating meaningful connections and understanding. The success of clinical trials heavily depends on how well we communicate and engage with patients. Here's what research has consistently shown works: Clear Communication of Value: ✅ Share how their involvement contributes to medical advancement ✅ Explain the scientific importance of their participation ✅ Highlight the structured nature of research protocols Building Competency: ✅ Offer detailed explanations of what to expect ✅ Create accessible resources for ongoing reference ✅ Use plain language to explain complex procedures ✅ Provide comprehensive education about trial processes Fostering Relatedness: ✅ Create supportive environments for questions ✅ Establish consistent communication channels ✅ Maintain regular check-ins throughout the trial ✅ Address concerns promptly and thoroughly Practical Support Measures: ✅ Provide clear scheduling information ✅ Offer practical assistance with logistics ✅ Ensure easy access to trial coordinators ✅ Implement reminder systems for appointments The key is maintaining professional boundaries while delivering exceptional support. This means: - Supporting informed decision-making - Avoiding overly emotional appeals - Focusing on factual information - Maintaining scientific integrity When we approach patient engagement through these structured methods, we see improved retention rates and more reliable trial outcomes. The focus should remain on facilitating understanding and supporting participation through practical, well-designed systems. #ClinicalTrials #PatientEngagement #ClinicalResearch #TrialRetention #ResearchBestPractices #PatientSupport #ClinicalTrialAwareness

AI & Real-World Data: Transforming Clinical Trial Recruitment. Clinical trial recruitment remains one of the largest barriers to delivering new therapies to patients. AI and real-world data (RWD) are transforming this process — enabling faster identification, better matching, and more inclusive enrollment across therapeutic areas. Key AI opportunities. - AI-powered patient identification – Advanced algorithms mine EHRs, registries, and genomic/lab datasets to find eligible patients in real time, even for complex biomarker-driven protocols, while improving diversity by identifying underrepresented populations. - Patient-centric engagement – AI navigators, chatbots, and personalized outreach guide patients and caregivers from trial discovery through eligibility verification, documentation, and site connection — offering 24/7 support to reduce drop-offs. - Site enablement – Automated pre-screening, point-of-care recruitment tools, and integrated diagnostic AI (e.g., endoscopy AI for IBD) cut manual workload, lower screen failure rates, and accelerate first-patient-in timelines. - Sponsor intelligence – RWD-driven feasibility and predictive analytics optimize protocol criteria, site selection, and enrollment targets; real-time monitoring enables proactive adjustments to keep timelines on track. Therapeutic Area Specific Opportunities. * Oncology – Rapid identification of biomarker-specific candidates from pathology/genomic reports; AI prompts at point-of-care improve referrals; targeted outreach drives diversity in trial participation. * Neuroscience – Predictive AI models forecast disease progression in Alzheimer’s and other CNS disorders, reducing high screen-failure rates and ensuring timely enrollment of patients most likely to benefit. * Immunology – Embedding AI into diagnostic workflows (e.g., colonoscopy scoring in IBD) identifies candidates during standard care; lab and imaging AI tools match patients with rare biomarker requirements. * Cardiovascular – AI processes data from wearables, remote sensors, and EHRs to identify and risk-stratify patients; decentralized trial models expand reach to rural and mobility-limited populations. * Rare diseases – AI harmonizes patient registry data globally to locate small, geographically dispersed populations, matching patients to highly specialized trials in record time. Global challenges in use of AI. Variability in data digitization, interoperability, privacy laws, and regulatory acceptance requires flexible, region-specific AI strategies to remain compliant and effective. At Thermo Fisher Scientific’s PPD clinical research business, we’re delivering these innovations today. Our Patient First digital solutions and TrialMed™ platform integrate AI-enabled patient recruitment, global site networks, and home trial services to bring trials directly to patients, reduce site burden, and meet or exceed enrollment timelines — accelerating life-saving innovation delivery worldwide.

You’re Missing Something Big. It’s Killing Your Enrollment. Screen Failure Logs are lying to you. Here’s the truth They won’t reveal all the reasons your patient enrollment is tanking. That’s terrifying. These aren’t just any reasons. They’re the Silent Enrollment Killers—the ones that sneak in and sabotage your study while you’re busy looking elsewhere. I call them silent because unless you know where to look, you’ll never hear about them. They’ll derail your study quietly, and you won’t even realize it until it’s too late. Let’s break down these covert killers stopping your sites from enrolling patients. 𝟭. 𝗦𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗠𝗼𝗻𝗶𝘁𝗼𝗿 𝗥𝗲𝗹𝗮𝘁𝗶𝗼𝗻𝘀𝗵𝗶𝗽𝘀 Imagine you’re the site staff, and every interaction with the monitor feels like dealing with the ‘study police.’ They’re unhelpful, or worse, they don’t know what they’re talking about, but they still pile on extra work. 𝟮. 𝗡𝗼 𝗠𝗼𝗻𝗶𝘁𝗼𝗿 𝗥𝗲𝗹𝗮𝘁𝗶𝗼𝗻𝘀𝗵𝗶𝗽 Now, take it a step further—what if the monitors are practically invisible? The message to the site? “You’re just a patient factory to us.” Guess which study gets deprioritised when there’s a better relationship elsewhere? 𝟯. 𝗦𝘁𝘂𝗱𝘆 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 𝗧𝗵𝗮𝘁 𝗔𝗿𝗲 𝗮 𝗡𝗶𝗴𝗵𝘁𝗺𝗮𝗿𝗲 You’d think making things easier on the sponsor’s end would help, right? Wrong. If it means dumping all that complexity onto the site, they’re going to start prioritising ‘easier’ studies. Your trial? Bottom of the list. 𝟰. 𝗜𝗴𝗻𝗼𝗿𝗲𝗱 𝗦𝗶𝘁𝗲 𝗙𝗲𝗲𝗱𝗯𝗮𝗰𝗸 Sites are on the front lines—they know why patients aren’t enrolling. But if you ignore their feedback and keep pushing them to enroll anyway, they’ll mentally check out. Nothing is as frustrating as addressing the issue six months too late. 𝟱. 𝗟𝗼𝘄 𝗦𝘁𝘂𝗱𝘆 𝗥𝗲𝗶𝗺𝗯𝘂𝗿𝘀𝗲𝗺𝗲𝗻𝘁 If your study is tough and it doesn’t pay as well as others, guess what? Your trial gets bumped to the back of the line. Sites have limited resources, and they’ll allocate them where it makes the most sense, financially and logistically. 𝗛𝗼𝘄 𝘁𝗼 𝗗𝗲𝗳𝗲𝗮𝘁 𝗦𝗶𝗹𝗲𝗻𝘁 𝗘𝗻𝗿𝗼𝗹𝗹𝗺𝗲𝗻𝘁 𝗞𝗶𝗹𝗹𝗲𝗿𝘀 𝗣𝗮𝘁𝘁𝗲𝗿𝗻 𝗥𝗲𝗰𝗼𝗴𝗻𝗶𝘁𝗶𝗼𝗻: Non enrolling sites? Don’t just sit there—call the study nurses and coordinators directly. 𝗚𝗲𝘁 𝗜𝗻𝘃𝗼𝗹𝘃𝗲𝗱: If you’re working with a CRO, respect their processes but insist on being present during site interviews. You need to hear the unfiltered truth. 𝗔𝗰𝘁 𝗙𝗮𝘀𝘁: As soon as you identify the problem, move quickly to fix it. And keep the sites updated on your progress. 𝗦𝗵𝗼𝘄 𝗨𝗽: There’s no better way to show a site they’re valued than by visiting in person. It matters. 𝗙𝗼𝗹𝗹𝗼𝘄 𝗧𝗵𝗿𝗼𝘂𝗴𝗵: After you’ve addressed the issue, check back in to ensure it’s really resolved. Don’t assume everything’s fine just because you fixed one problem. 𝗥𝗲𝗺𝗲𝗺𝗯𝗲𝗿 Enrollment isn’t just about patients - your sites matter just as much! P.S. Which Silent Killer do you see the most (1-5)? Comment below.

Half the battle of patient recruitment is getting a potential participant to respond to your outreach. This challenge is particularly acute in the US. Americans receive an average of 26 scam calls and 11 scam texts a week. People have learned to ignore unknown numbers entirely. Your legitimate recruitment outreach gets lumped in with car warranty ruses and fake IRS threats. Let’s compare a cohort of American sites to similarly sized European peers. Both cohorts recruiting for very similar studies. Both using a combination of SMS and phone outreach from within their existing database. Here’s what our data shows us…  • US sites see a 20% lower pick up rate on their first phone attempt  • After three call attempts this gap has grown to 28% 🤕 Ouch! So, how can you improve reach rates? Short of transplanting your site to a different country we’ve found there to be four key levers you have to improve your engagement…  • 𝗥𝗲𝗹𝗲𝘃𝗮𝗻𝗰𝗲: While your participant database is your greatest asset for recruitment, leads generated from advertising are 15% more likely to engage when compared to an older database lead.  • 𝗕𝗿𝗮𝗻𝗱𝗶𝗻𝗴: Implementing call branding for our customers resulted in meaningful lifts in the SMS response and call pick up rates. If you ain’t brandin’ you ain’t landin’.  • 𝗖𝗵𝗮𝗻𝗻𝗲𝗹: Combining phone calls, SMS and email together provides participants with greater certainty that you are legitimate. Sending an SMS prior to a phone call improves pick up rates by 5%.  • 𝗖𝗼𝗻𝘀𝗶𝘀𝘁𝗲𝗻𝗰𝘆: The average lift in pick ups between a first call and second call attempt is 10%. For later contact attempts this diminishes but is consistently around 2-3%. Combining all of these levers together levels the playing field. For US sites following these best practices, we see that our blended approach across multiple channels and with 7+ contact points, is consistently driving reach rates above 75%.