The contents of this article are informational only and are not intended to be a substitute for professional medical advice, diagnosis, or treatment recommendations. This editorial presents the views and experiences of the author and does not reflect the opinions or recommendations of the publisher of Ophthalmology 360.
By James Fox, MD, and Ram Yalamanchili
Clinical research sites are facing a mounting operational crisis. While the number of trials continues to grow and protocols become more complex, the systems and staffing models supporting site operations have remained largely static. This creates an increasingly unsustainable gap between sponsor expectations and site capacity. For investigators on the ground, it’s a bottleneck that can affect trial performance, patient access, and staff wellbeing.
At ICON Eyecare in Grand Junction, Colorado, our ophthalmic research program has grown steadily over the last 9 years. We’ve expanded to manage over 25 active trials with a hybrid staff of three and a half research coordinators. But despite our clinical and operational success, our team faces the same limitations shared by many research sites: unpredictable study volumes, labor-intensive workflows, and an inability to scale without overworking our most trusted staff.
The root of the problem is not the research itself, which continues to evolve. It’s the infrastructure supporting research, particularly at the site level, that has failed to modernize.
The Scalability Ceiling in Site Operations
Sites operate in cycles: study startups, recruitment peaks, enrollment slowdowns, and data closeout. In a program like ours, we often have 10 to 15 studies recruiting simultaneously. When multiple studies with broad inclusion criteria ramp up in parallel, it creates a surge of work that even our most efficient coordinators struggle to keep up with.
Staffing against this kind of variability is a perennial challenge. As we’ve experienced firsthand, moving from 1 to 2 full-time coordinators is a leap; moving from 2 to 3 introduces exponential complexity in training, oversight, and quality assurance. Because clinical trials are built on trust between sponsors and sites, PIs and their teams, we find that the quality and consistency of site personnel matter as much as the actual headcount.
Beyond recruitment, a primary concern for research sites is also the retention of qualified staff. A 2023 survey by the Association of Clinical Research Professionals (ACRP) revealed that 63% of research sites identified staffing and retention as their top challenge.1 The shortage of skilled personnel not only delays study initiation but also affects the quality of data collected. Moreover, high turnover rates can lead to increased training costs and loss of institutional knowledge.
For a real-world example, consider that at one point, one of our coordinators logged 20 hours of overtime weekly before asking for relief. We recognized the danger signs: looming burnout, data lag, and the growing temptation to turn down trials we were otherwise well-positioned to conduct. What we lacked was a way to expand operational capacity without overextending our people.
Patient Recruitment and Enrollment
Closely following staffing issues, patient recruitment and enrollment pose significant hurdles. The same ACRP survey indicated that 48% of sites struggle with enrolling sufficient participants. Factors contributing to this challenge include stringent eligibility criteria, patient apprehension, and lack of awareness about clinical trials. These issues can lead to extended timelines and increased costs.
Complexity of Clinical Trials
The increasing complexity of trial protocols adds another layer of difficulty. Adaptive designs, multiple endpoints, and intricate procedures require meticulous coordination and expertise. This complexity can overwhelm site staff, leading to protocol deviations and data inconsistencies.2,3 Additionally, complex trials often necessitate advanced technological infrastructure, which may not be readily available at all sites.
Technological Integration
The integration of various technological systems, such as Electronic Data Capture (EDC), Clinical Trial Management Systems, and Electronic Health Records (EHR), is essential for efficient data management. However, interoperability issues among these systems can lead to data silos and inefficiencies. For instance, challenges in integrating eSource data with EDC systems can result in duplicate data entry and increased potential for errors. What’s more, site staff are already overburdened with poorly functioning technology and are often not keen on adding more tools, which require separate training, log-ins, and other distractions.
Regulatory Compliance
Navigating the regulatory landscape is inherently complex. Sites must manage a plethora of regulatory documents, including protocols, investigator brochures, informed consent forms, and case report forms. Ensuring compliance with Good Clinical Practice guidelines and maintaining accurate documentation requires substantial effort and resources. Non-compliance can lead to severe consequences, including study termination and legal ramifications.
For research sites, these are just some of the elements that often impede rather than facilitate their focus on the clinical study itself.
Early AI Adoption: A Practical Use Case
For these reasons, and more, we began seeking out artificial intelligence (AI) teammates to help handle core research workflows such as regulatory document prep, data entry, and financial reconciliation. From the outset, we understood this wasn’t about replacing our coordinators. It was about freeing them from repetitive, time-intensive tasks.
Within the first 3 months of implementation, the operational benefits were immediate:
- Data entry time reduced by 47% across all studies using AI support
- Query response times improved by 31%, reducing delays in data cleaning and review
- Manual queries per site dropped by 42%, largely due to preemptive AI validation
- Study startup processes accelerated significantly: templates, reg binders, and delegation logs prepared in hours instead of days
Perhaps most importantly, these efficiencies allowed our team to train a new part-time coordinator, something we would not have had the capacity to do otherwise.
Human + AI Teammate = A Better Site Model
Research remains deeply human. Trust, judgment, and adaptability are traits no algorithm can replicate. That’s precisely why the current model is broken. We are asking our most capable people to spend time on tasks that don’t require their clinical judgment or site expertise.
Prior to AI integration, we experimented with temporary staff to manage the workload. But temps lack the domain context and consistency required in regulated research. What AI teammates enabled us to do is something entirely new: the ability to support the core site team with consistent, domain-trained assistance that improves over time.
As Dr. Houman Hemmati has argued in his sponsor-side perspective, the burden placed on sites has become one of the most significant rate-limiting factors in clinical research today.4 From the sponsor’s perspective, delays in site activation, missed enrollment milestones, and inconsistent data quality all tie back to operational fragility at the site level.
Similarly, in a coauthored work by Dr. George Magrath of Opus Genetics, AI is positioned not simply as an efficiency tool but as a strategic asset to expand access to research by enabling more sites to participate, especially those with limited infrastructure.5 Our experience supports this thesis. AI does not replace coordinators; it makes it more feasible for small- to mid-sized sites to operate at the level of larger, resource-rich centers.
Building Toward a Scalable Future
We’ve now seen that AI teammates can functionally act as one and a half to 2 additional study coordinators, scaling up or down dynamically depending on study load. This flexibility is something traditional hiring models simply can’t offer. Our goal with AI teammates is to give our experts room to operate at the top of their license.
The next phase of impact, in our view, lies in accelerating study startup timelines. From our view, we are moving toward a future where study activation processes can be completed within 24 hours of execution rather than the traditional weeks. That level of readiness is especially important in ophthalmology, where studies can fill quickly and inclusion windows are short.
We are also beginning to explore how AI can support patient matching by screening structured and unstructured EHR data in real-time. In high-throughput programs like ours, where recruitment conversations often happen face-to-face in the clinic, real-time insights could dramatically improve both enrollment and patient access to novel therapies.
Conclusion: A Site-Centered AI Revolution
The promise of AI in clinical trials isn’t about automation for its own sake. It’s about creating breathing room for coordinators, investigators, and research itself. We believe sites like ours, working in close collaboration with AI developers and sponsors, can lead the way toward a more scalable, equitable research ecosystem.
This is not just about our research site or even just ophthalmology. The challenges we face are systemic. But they are solvable. With the right tools and partners, we can finally start dismantling the operational brick walls that have kept site capacity lagging behind trial demand.
If AI can help us do that, then it belongs not in the future of clinical research, but in its present.
James Fox, MD, serves as the Western Region Medical Director of ICON Eyecare in Grand Junction, Colorado. Dr. Fox specializes in both medical and surgical treatments for patients with cataracts, glaucoma, intraocular lens complications, trauma, and those who might benefit from refractive surgery.
Ram Yalamanchili is the CEO and co-founder of Tilda Research and a former Clinical Research Coordinator. He previously founded Lexent Bio and believes the future of clinical research is through augmentation with AI teammates.
Dr. Fox and Ram Yalamanchili have nothing to disclose.
References
- Association of Clinical Research Professionals. Top site challenges of 2023 – data and insights on site burden and trial efficiency. August 2, 2023. Accessed June 3, 2025. https://acrpnet.org/2023/08/02/top-site-challenges-of-2023-data-and-insights-on-site-burden-and-trial-efficiency-2
- Getz KA, Smith Z, Jain A, Krauss R. Benchmarking protocol deviations and their variation by major disease categories. Ther Innov Regul Sci. 2022;56(4):632-636. doi:10.1007/s43441-022-00401-4
- Burnett T, Mozgunov P, Pallmann P, Villar SS, Wheeler GM, Jaki T. Adding flexibility to clinical trial designs: an example-based guide to the practical use of adaptive designs. BMC Med. 2020;18(1):352. doi:10.1186/s12916-020-01808-2
- Hemmati H, Yalamanchili, R. Why clinical trials fail. YouTube. May 14, 2025. Accessed June 3, 2025. https://www.youtube.com/watch?v=v-aikU6EnKs
- Yalamanchili R, Magrath G. Transforming clinical research: the economics of clinical research in the era of AI [unpublished manuscript]. 2025.
Related Content
