Some patients qualify for a clinical trial and never hear about it. For Dr. Arturo Loaiza-Bonilla, that problem began in a University of Miami clinic.
He treated a largely uninsured, underrepresented population there. Many patients were trial-eligible but invisible to the systems meant to find them.
The Problem Medicine Couldn’t Fix
“There was nothing that I could do for those patients,” he said. “But I said, ‘we need to do something.’”
The problem followed him to Penn Medicine. Even colleagues in the same department did not know which studies were open.
He was quick to say it was not a lack of care. Incentives were not aligned, and the time was not there.
So he left academia. He co-founded Massive Bio and set out to build the infrastructure oncology was missing.
Today he is system-wide chief of hematology and oncology at St. Luke’s. Cancer Network profiled him after a recent episode of The Moonlight Shift.
In the profile he explains how AI is matching patients to trials. He also covers where the technology falls short.
A patient who never hears about a study cannot join it. He saw that gap firsthand, before he ever wrote a line of product code.
Massive Bio’s AI platform characterizes a patient and matches them to trials end to end. The company says that can take as little as 17 seconds.
It pulls structured data from US electronic medical records. Then it cross-references a fully parameterized clinical trial database.
That database was developed in collaboration with OpenAI. The goal is to replace a slow manual search with a fast, repeatable one.
The company says the platform has helped more than 200,000 patients across 17 countries. It has earned an NCI SBIR grant and published a prospective evaluation in 3,800 patients.
Those are company figures. They show reach, not proof that patients enrolled.
Where the Technology Falls Short
Loaiza-Bonilla is candid about limits. He was asked whether AI can fix inequity in trial access, and his answer was measured.
Alert-based matching systems, including one randomized at Harvard, produced more matches. They produced no improvement in enrollment.
The gap, he said, is not identification. It is engagement, education and the last mile.
In other words, finding a trial is the easy part. Getting a patient to say yes, and to show up, is the hard part.
A patient must understand the trial, trust it and reach the site. Software can surface the option but cannot make that walk.
On whether AI will make cancer care more equitable, he said, “Jury still out.” He added, “We can do both.”
A Miami Thread in Cancer AI
The story starts in Miami and still bears on it. South Florida’s cancer centers are already weighing AI at the bedside.
MAIN reported that AI enters the tumor board, and Miami Cancer Institute is in.
Better data is also a theme here. MAIN covered why better healthcare AI starts with better clinical data.
Miami’s health AI firms are working on adjacent problems. MAIN profiled five Miami health AI firms beyond the clinic.
Loaiza-Bonilla’s point applies to all of them. Finding the right patient is a data problem, but getting care to them is a human one.
His clinic’s patients are the reminder. People who are uninsured or underrepresented can be eligible on paper and absent in practice.
For Miami’s hospitals and startups, the lesson is practical. Build the matching tools, then fund the people who close the last mile.
Sources:
- Cancer Network: Can AI Fix Clinical Trial Access in Oncology? Dr. Arturo Loaiza-Bonilla Built the Answer With the Data to Back It
- MAIN: AI Enters the Tumor Board. Miami Cancer Institute Is In
- MAIN: Better Healthcare AI Starts With Better Clinical Data
- MAIN: 5 Miami Health AI Firms Beyond the Clinic