Intermountain Health, the largest nonprofit healthcare provider in the Intermountain West region, has worked with a young AI startup this month for many years.
The health system is working with Boston-based healthcare AI company Lays Health, which uses large language models (LLMS) to improve clinical data abstraction without increasing administrative burden. Intermountain’s venture capital arm is making strategic investments in the startup, and the organization is deploying technology in several of its clinical registries.
Clinical data abstraction is the process of manually reviewing and extracting critical information from patient medical records – which often means taking unstructured data and converting it into standardized available data. CEO David Sontag explained that the layer uses LLM and a range of other machine learning techniques to automate the process.
“For clinical registry abstraction use cases, we fine-tune a range of AI models that can reason in the patient’s records and answer questions found in the clinical registry, such as “Is thrombolysis performed for this patient’s stroke? ” and “If not, what are the taboos? “This AI-enabled chart review engine is used to power the software used by Intermountain’s clinical abstraction team – the software provides suggested answers to clinical registry questions and has received a lot of evidence support in patient charts.”
The scope of the partnership includes registries for cardiovascular, stroke and bariatric surgery spaces.
Sontag notes that for each registry, a complete longitudinal patient chart is often required to be parsed, including structured data elements such as laboratory, vitality and diagnostic codes, as well as unstructured data such as clinical annotations and radiological reports.
He noted that Intermountain participated in a large number of clinical registries, which required a lot of manual manual chart abstraction work.
“Using Layer’s software, Intermountain has the opportunity to enable its high-quality employees to ‘license’ the highest license, which is based on operational improvements based on data insights, rather than spending most of its time collecting data.”
Phillip Wood, executive director of Intermountain Ventures Partnership, said his team has been focusing on solutions to improve clinical data abstraction and when they met with Layer’s team, they thought the technology was suitable for their organization.
“We were impressed by our team and approach when we were introduced to Layer Health and were excited by the early success they achieved in their initial marketing. After a technical demonstration with our team of clinical excellence, we determined that Layer would be a good partner to work with to develop and validate the use of this AI technology,” Wood said.
He added that to determine the success of the partnership, Intermountain will measure the efficiency and accuracy of the platform compared to previous benchmarks.
“If we can improve efficiency and accuracy with tools, this will enable our caregivers to do more with other quality organizations and improvement programs,” Wood said.
Photo: ra2studio, Getty Images
