PI's CRAACO Model + AI = Rapid Enrollment

By Dan Schell, Chief Editor, Clinical Leader

It’s weird that I first learned about — and met — David Almeida, MD, MBA, PhD at the MAGI 2025 conference in Boston considering he’s the PI (and president & CEO) at Erie Retina Research, a very active clinical trial site in my hometown of Erie, PA. His MAGI presentation discussed how his proprietary AI model and his BLOC enrollment strategy significantly accelerate patient enrollment at his site. I’m talking a screen-to-enroll percentage of 85 in an average of 2-3 months for the eye-related trials he’s involved with (Almeida is an ophthalmologist). Those metrics attracted a lot of attention, but so did statements like this:

“As of January 1, 2024, we follow the CRAACO model [Clinical Research As A Care Option]. I don’t take insurance; we are not even set up to take insurance. We don’t take any money from patients; we get paid from the pharmaceutical companies that we do trials for. So, since I’m responsible for paying for the care of anyone who screen fails, it really makes me think about ways to increase my screen-to-enrollment conversion. That’s where this BLOC enrollment strategy comes in.”

When I returned to Erie, I took him up on his offer to visit his site. When I arrived, the place was buzzing with activity in the halls and behind the reception desk as numerous (there are about 30 employees) staff cycled patients in and out of the almost packed waiting room. Absent were the typical signs about insurance cards or payment plans that often adorn the walls and glass partitions of physicians’ offices. Instead, I noted only two sheets of paper posted: one for a REGENXBIO ATMOSPHERE clinical study and another that said “Blazing new trials in clinical research.”

Considering they’re currently managing around 50 active studies with more than 1,000 patients enrolled, I could tell this place wasn’t moonlighting as a research site.

EMBRACING INNOVATION

Almeida’s wife Jasmine, who is the company’s director of communications and marketing/director of patient recruitment and referrals, was my host at the office, answering all my questions as her husband would stop by in-between seeing patients. She explained that Dr. Almeida has made a name for himself in retina circles and is a regular speaker at major ophthalmology, and now, some clinical trial-related conferences. As he is typically one of the first physicians to enroll patients in new trials, that reputation attracts sponsors at those shows, so finding trials is often not a problem. They employ eight coordinators and have on-staff grant writers to supplement initiatives like a mobile eye-screening unit that they can take into the community. 

Innovation seems baked into the clinic’s DNA. Jasmine explained how she designed a HIPAA-compliant referral app, which allows local ophthalmologists to submit patient referrals directly into the clinic’s trial workflows — including scheduling, diagnostics, and imaging. On top of that, Dr. Almeida has created an in-house AI tool for study coordinators that enables them to do things such as query the protocol — “Can this patient on antidepressants enroll?” “What’s the washout period?”— without flipping through a 40-page document. It also can upload documents and send queries to sponsors. They’re testing it now, but it's already functional and protocol-agnostic.

AN AMBITIOUS AI-DRIVEN PATIENT ENROLLMENT MODEL

There’s no doubt that much of Erie Retina Research’s success can be attributed to its BLOC enrollment strategy. Early on, Dr. Almeida realized that the traditional approach to enrollment — gradually accumulating the total number of patients needed over months or even years — was not only inefficient, it was expensive when employing the CRAACO model.

“I started pondering questions such as, ‘What if we could enroll all the patients we need for a trial in a week?’” he said. “So, for example, if the goal is 100 patients, what if we could enroll 20 patients each workday so that by the end of the week everyone’s enrolled?”

The concept sounds ambitious, but the Erie site has been running BLOCs — more than 30 of them so far — for years, and the results (the aforementioned 85% screen-to-enroll conversion rate) back up the theory.

At the heart of this model is a multistep, AI-driven process which starts with a “grand sorter” as Almeida describes it. Once the sponsor makes a request for X number of patients for X disease state, the AI cross-references the site’s EMR database and flags potential matches. Then, selected patients are invited in for a free visit with Dr. Almeida often personally seeing them. Metrics from that live screening visit are reviewed again, this time through a second algorithm that mimics the sponsor’s reading center (i.e., a provider of unbiased, independent image analysis). Finally, the case is evaluated by the reading center itself.

The process isn’t just efficient — it’s predictive. “Our ability to predict which patient goes in which trial gets really good because we’ve analyzed all of this data,” Dr. Almeida says.

Each BLOC involves a full-team commitment to one protocol. That might mean concentrating on just one trial for multiple days, but the outcome is immediate enrollment volume. The trial doesn’t start at zero. It starts with 10, 20, even 30 patients already enrolled and ready to go.

REAL COST SAVINGS

For sponsors and CROs, the benefits are obvious. Condensing trial enrollment reduces site monitoring time, data management cycles, and overall trial duration — which translates directly into cost savings. “When you think about CRO costs and reading center costs and all the costs to do research, if you can shorten the trial by three months, you now have real cost savings,” Dr Almeida said.

But cost savings aren’t the only benefit. Erie’s team found that quality and safety actually improved under the BLOC model. Because all the patients were going through the same review process at the same time — using the same equipment, evaluators, and environmental factors — variability was reduced. A quality review happens on the same day as the assessment. Data is entered directly into the site’s eSource application and pulled into the sponsor’s EDC platform in real time.

Sponsors have taken notice; some have started asking for BLOCs by default — requesting monthly enrollment blocks rather than rolling recruitment. In a global Phase 3 trial, Erie enrolled 9% of the total patient population — from one site, with one PI. All through BLOCs.

Dr. Almeida’s model is not easily copied. It requires infrastructure, vision, and staff buy-in. It also demands a level of operational discipline that many sites — especially those juggling competing clinical responsibilities — simply can’t achieve.

But it’s also proof of concept. Clinical trials, when treated as a primary care option, can not only support an entire clinic — they can lead it.