Boston Children's says AI has unlocked 40-plus rare disease diagnoses

Boston Children's Hospital says artificial intelligence has helped its clinicians diagnose more than 40 rare conditions that had previously gone unresolved, alongside capturing roughly 60,000 hours of staff time savings — more than $7 million in redeployed labor — across more than 50 automations.

The numbers, disclosed by the hospital's innovation leadership, describe one of the most concrete deployments of AI inside a major US health system. Boston Children's, one of the largest pediatric institutions in the world, serves patients across more than 40 specialties and handles close to 1 million outpatient visits each year. The stakes are significant for a sector squeezed by tight financial constraints, rising administrative burden and clinical problems — such as rare disease diagnosis — that exceed what physicians can process unaided.

The bottleneck is cognitive, not motivational

Rare disease cases typically involve fragmented genetic data, incomplete clinical histories and a medical literature too vast for any physician to synthesize in real time. That limitation applies even at a leading research institution.

"The problem isn't effort," says John Brownstein, Chief Innovation Officer at Boston Children's. "It's human cognitive limits."

The operational side faces a parallel constraint. Teams across supply chain, billing and operations process high volumes of repetitive tasks — invoices, scheduling coordination — that pull staff away from higher-value work.

From one-off tools to an enterprise AI layer

Boston Children's did not start at scale. Early efforts involved individual AI use cases, including documentation and translation tools. Those pilots exposed the limits of a fragmented approach.

"You cannot just rely on one-off solutions," Brownstein says.

The hospital shifted to what Brownstein calls an enterprise AI layer: a secure internal ChatGPT environment used across research, clinical and administrative teams. Rather than treating AI as a collection of disconnected tools, the organization built a shared foundation on which new capabilities can be developed and deployed quickly. Governance structures for safety, monitoring and consistent evaluation were built alongside the technology.

The pace of deployment changed as a result. Tools that once required extended development cycles can now go live in days. Today, more than one-third of employees use AI daily, spanning clinical, research and administrative functions.

Redesigned workflows, measurable returns

The hospital targeted areas where AI could deliver measurable operational impact first. In supply chain operations, AI now manages invoice intake, routing and responses. In surgical scheduling, the system analyzes clinical notes and estimates patient acuity to improve how operating room time is allocated — allowing schedules to be planned further in advance, increasing utilization and getting patients into care faster.

Physicians use AI for decision support and to synthesize complex clinical information. Researchers apply it to data analysis and cohort building. Administrative teams use it for drafting documents, coding and workflow improvements.

The organization ties these changes directly to hard numbers: across more than 50 automations, roughly 60,000 hours of time savings, equivalent to more than $7 million in redeployed labor.

Brownstein attributes adoption to relevance rather than mandate. "The key here is meeting people where they are," he says.

A 'co-pilot geneticist' for unsolvable cases

The clinical discovery work is the most notable result. Boston Children's developed what it describes as a "co-pilot geneticist" — a system that integrates genetic data, phenotypic information and global medical literature.

The system targets one of medicine's hardest problems: rare diseases that have eluded explanation for years. To date, it has produced more than 40 diagnoses previously thought impossible, and the hospital says the work has also identified new gene targets and potential therapeutic pathways.

"We combine genetic information, phenotypic information, literature search, and the reasoning of AI to deliver diagnoses to families that were once left without any answers," Brownstein says.

For families, the hospital frames the impact in immediate terms: previously unresolved cases are now yielding answers and, in some cases, new treatment directions.

"This was unthinkable before, but is now providing hope to so many families," Brownstein says.

What comes next

The next phase of the hospital's AI strategy focuses on deeper integration into clinical decision-making, extending tools across specialties, and refining models through collaboration with OpenAI. Leadership expects AI to become a core component of medical practice over time.

"How would you not want an incredibly trained physician alongside all the world's medical knowledge?" Brownstein said.

For other health systems weighing AI investments, Boston Children's offers a template: a shared, governed enterprise layer rather than scattered pilots, tied from the start to measurable operational savings and clinical outcomes. The hospital's stated trajectory — from documentation tools to a diagnostic system producing previously impossible rare disease diagnoses — indicates where it expects the technology to go next.