White Plains Hospital is teaming up with health tech company Layer Health to use its artificial intelligence platform to automate clinical registry reporting, a highly manual process that typically requires significant manpower.
Clinical registries are structured databases that track outcomes for patients with specific conditions or procedures. They play a key role in improving care quality, supporting research and guiding best practices in real-world settings.
White Plains Hospital, located in Westchester County just north of New York City, participates in numerous clinical registries, each of which requires significant resources to collect and submit high-quality patient data. The exhaustive, largely manual process of chart review demands substantial time from clinicians, nurses and staff, who dedicate hundreds of hours each year to analyzing health records.
"The organization has been struggling with this issue for some time now. We have 15 clinical registries that we participate in, and 25 FTEs [full-time equivalents] at the moment," Rafael Torres, M.D., chief quality officer at White Plains Hospital, told Fierce Healthcare.
White Plains Hospital, which is a part of the Montefiore Health System, is on a strong growth trajectory as emergency department volume has nearly doubled and inpatient volume growth has kept pace, Torres noted.
"That trajectory is only onward and upward for these clinical registries, and we're going to need more people and more highly trained people in order to keep up with them as this is highly skilled work," he said.
The 292-bed hospital serves as Montefiore's tertiary hub of advanced care in the Hudson Valley and operates outpatient medical facilities and multispecialty practices across Westchester County as well as Scarsdale Medical Group locations.
White Plains Hospital has received a five-star rating from the Centers for Medicare & Medicaid Services (CMS) for three consecutive years. The hospital also received its third Magnet designation from the American Nurses Credentialing Center and has earned an “A” Safety Grade from the Leapfrog Group for more than 10 years.
"We want to maintain or improve upon our 5-star rating that we have by CMS, our Leapfrog 'A' grade, our Magnet status. In order to be able to do that, we need the right people and we need to support the right people with the right technology," Torres said.
Layer Health’s approach to AI-enabled chart review and clinical registry reporting will enable the hospital to meet and exceed quality reporting requirements while freeing up its team to focus on meaningful clinical improvements, Torres said. Layer Health's AI-driven abstraction process will help the provider scale quality initiatives across the organization, he noted.
"We know they have a deep bench. We know that the time to value is going to be very short because they have a product that is ready to go," Torres added.
An AI company out of the Massachusetts Institute of Technology, Layer Health brings the first comprehensive chart review algorithm to market. David Sontag, Ph.D., CEO of Layer Health and a professor at MIT, founded the company with Divya Gopinath, Luke Murray, Monica Agrawal, Ph.D., and Harvard emergency physician Steven Horng, M.D.
Clinical registries and chart abstraction are key to understanding what treatments work best for patients and to standardizing clinical care, Sontag said. To answer the long list of questions thoroughly, Layer has trained its algorithm to be able to look at a much larger swath of unstructured patient data than other large language models of its kind, according to the company.
Sontag and his co-founders launched Layer Health with a focus on rethinking the way providers, payers and patients interact with patients' medical records and took aim at one of the biggest pain points in healthcare: chart review, Sontag said.
"Everywhere across healthcare, we have physicians, coders and nurses reading the patient's medical records to try to piece together a story of what's going on with that patient. That, of course, drives clinical decision-making, but it also drives a huge amount of administrative burden in healthcare," he said in an interview.
Submitting data to clinical registries often involves highly trained nurses spending an hour or more per patient reading the medical record to answer a long list of questions about the patient, Sontag noted.
"The challenge here is that these questions are clinically very nuanced, and to answer one of these questions, the nurse is typically clicking around through the electronic health record, reading through lots of the notes written by the different providers for the patient, trying to reconcile what the notes are saying against a lot of the structured data you have, the medication entries, and there's often a lot of conflicting information as well," Sontag said.
"What Layer Health is building is AI for chart review which can read through a patient's medical record and understand a patient's journey just as well as the best care team that has infinite time to read that patient's medical record and then use that to really just drive downstream decision-making and downstream administrative efficiencies," Sontag said.
Layer Health's AI platform can read through the patient's medical record and "tees up" the answers for the nurses and provides evidence to support the answers from the medical record, he added. The company's technology enables hospitals to scale their registry reporting without the need for additional staffing.
Layer Health picked up $21 million in series A funding in March backed by Define Ventures, Flare Capital Partners, GV and MultiCare Capital Partners. MultiCare Health's venture arm also has invested in the startup.
Chart review and quality reporting is an area ripe for automation and AI, and many startups have jumped into the sector. Brellium built an AI-powered solution that helps providers automate clinical quality and payer compliance and recently raised $16.7 million. Dyania Health, which uses AI to automate manual patient chart review, picked up $10 million in series A funding last fall.
How White Plains Hospital and Layer Health are build out AI for chart review
Sontag noted that White Plains Hospital is among the more "forward-thinking" health systems with adopting AI. "They are very ambitious. This is not a small pilot that we're tackling. It's a multi-year partnership which aims to take a very big bite out of these manual efforts and really enable them to scale."
White Plains Hospital was looking for a partnership, not just a vendor relationship, to build AI solutions that support the work that staff and care teams are currently doing, Torres said.
"This is not just our taking our technology that's pre-baked and handing it to them, but really working very closely with them to make sure it's addressing their pain point and iterating as necessary to continue to make sure it addresses their pain point longer term. And it's one where I see a lot of opportunity to continue to grow," Sontag said.
Layer Health built its AI platform to use advanced LLMs trained on longitudinal patient data to enable health systems to automatically review and interpret both structured and unstructured clinical data at scale with clinician-level accuracy.
The company developed modules to automate clinical registry data submission to support all major registries including national surgery, cardiovascular and oncology registries.
Layer Health's technology helps "democratize" clinical registries, quality improvement, quality assurance and performance improvement, Torres noted, as it allows organizations to read through medical charts more efficiently, find opportunities for improvement and intervene sooner. "This allows clinicians and nurses to focus on care rather than focus on hunting and pecking through a chart," he said.
A key benefit to participating in clinical registries is to identify areas where the organization can improve care and safety, Torres said. "We like to measure and intervene, or something called a 'measure-vention' as quickly as possible. Because these registries take so long to complete, and there's so many of them, at times we we may not get to these records for weeks, if not a couple months, which delays the feedback to clinicians, to physicians and to nurses, and it delays performance improvement. The closer we can stay to care at any given time with these clinical registries, the better insights we'll have with the care that we're providing, and the sooner we can intervene," he said.
Layer Health was selected due to its academic rigor, proven accuracy and potential to enhance outcomes by reducing administrative burdens, according to Torres.
A key differentiator was Layer Health’s deep retrospective validation before go-live as the company used several years of historical data to calibrate the AI models and quantify model accuracy. This process builds trust with abstractors and ensures that the AI-driven, abstraction process meets the hospital’s high-quality standards before full implementation, he noted.
Other use cases for Layer Health's platform include clinical research and real-world data abstraction, hospital operations and revenue cycle management and clinical decision-making and patient care optimization. Froedtert & the Medical College of Wisconsin health network used Layer Health’s technology to streamline quality data abstraction, reducing the time required by more than 65%.