The Next Stop on Healthcare’s EHR Optimization Journey: AI
It’s been more than 15 years since the U.S. government passed the Health Information Technology for Economic and Clinical Health Act to incentivize healthcare organizations to adopt electronic health record systems (EHRs). Now, with the emergence of newer tools and processes, providers are looking to optimize these critical applications to streamline clinical workflows and hopefully address burnout.
“There’s a tremendous amount of energy happening with the adoption of artificial intelligence and machine learning that is pushing organizations to truly optimize the technologies around EHRs, with the focus on improving efficiencies, reducing burnout and enhancing patient care,” says Christopher Kunney, a health IT strategist who hosts the Straight Outta Health IT podcast.
Documentation is a key part of clinical workflows. It tends to be an administrative burden for providers, so organizations are turning to AI-powered solutions to offer better support. Natural language processing capabilities integrated with an EHR, for instance, can allow clinical note-taking in real time, then help to sift through notes with special keystrokes, he says.
In addition, machine learning models can now analyze EHR data to help health systems with precision medicine and predict diseases in populations, Kunney adds.
Despite the benefits that digitized systems can provide, clinicians are still not very excited to use them.
“You’re not going to have that love relationship with the EHR,” says Dr. Vinay Vaidya, senior vice president and chief medical information officer at Phoenix Children’s. “It’s just a tool.”
Working for 17 years with Phoenix Children’s, Vaidya has witnessed the evolution of the EHR within the large, Arizona-based pediatric organization, starting from a general system to one that includes more specialties, such as pulmonology and cardiology, he says. It all connects to a single EHR system that comprehensively incorporates inpatient and ambulatory areas and operating and emergency rooms.
Phoenix Children’s EHR journey consists of three phases: implementation, optimization and application, Vaidya adds. “Our third phase, which I think is the most important phase even before AI came into the picture, is bringing the EHR data back to all of the front-end people, the mid-level people, the leadership level. That has been a transformative experience at Phoenix Children’s in the past seven or eight years,” he says.
The discussion among clinicians has matured from focusing on the EHR to looking at what they can do with the data, such as researching how many patients in the health system have neurofibromatosis, a group of genetic disorders that can lead to tumor growth on nerves. Clinicians can then track this population closely to watch for early detection of malignant transformation, if any, Vaidya says.
READ MORE: Optimize your EHR and use data to improve clinical workflows.
Phoenix Children’s integrated a documentation tool in its central EHR to achieve structured data. Data flows into a homegrown, Microsoft-based data warehouse from the EHR in one-minute intervals, he notes. “We don’t have to wait for 24 hours. We’re operating on near real-time insights.”
More than 100 different electronic systems pull data into the data warehouse. The organization also uses Microsoft Power BI as its data visualization tool and is experimenting with connecting ChatGPT to the EHR platform using APIs “to summarize large volumes of clinical text and support ambient listening capabilities that can reduce documentation burdens,” Vaidya says.
Phoenix Children’s has begun using generative AI to help with chart summarization and EHR data analysis to predict malnutrition in kids.
“It was so successful that we put a policy in place that if the predictive model has identified malnutrition, we are automatically generating an order for the dietitians and the nutritionists to examine this patient without waiting for the physician,” Vaidya says.
The health system is also performing predictive analytics to detect sepsis risk earlier. These AI tools can provide “gentle nudges” to physicians rather than replacing them, he says.
Since West Plains, Mo.-based Ozarks Healthcare moved from paper to digital, optimizing the EHR has been ongoing, says Chief Medical Information Officer Dr. Priscilla A. Frase.
The challenge is balancing the joy of the job for physicians and nurses with the regulations, billing processes and need to create coherent documentation for patients, she says.
“I still don’t like EHRs. I avoided them like the plague early on,” Frase says. “But I discovered that I couldn’t overcome them, I couldn’t run away from them after a while, and I really needed to just embrace it and be a part of trying to make them better.”
This distaste for EHRs motivates Frase to help optimize workflows for herself and other clinicians. Through a partnered program, she and her team have identified areas the health system could address EHR pain points.
She also works with physicians to identify why they’re spending so much time doing EHR documentation. A deeper exploration into the platform, for example, led to the removal of duplicate requirements and updates to regulatory and reporting standards.
Now, her team has integrated ambient scribe documentation tools and is evaluating AI to help with chart summarization. Frase also led the build of a calculator to help the nursing staff manage dosing for heparin, an IV blood thinner for preventing blood clots; there are ongoing discussions about wider implementation of the calculator in the EHR’s platform.
“It’ll tell nurses exactly what they need to do, per our protocol, so they don’t have to do the math, which can take time and still include errors,” Frase explains.
As a small, rural health system, hiring talent while staying up to date on tech is a challenge. “We want to recruit for all healthcare staff and not have technology be yet another barrier to recruitment. It’s a problem if we’re not attracting talent because we don’t have the right tools,” she says.
The percentage of clinicians who say they need to work late or bring work home because of excessive documentation tasks
As patients increasingly rely on health apps and online portals, their tolerance for completing paper forms has decreased, which has been a motivating factor to optimize EHRs, says Julie Demaree, executive director of clinical innovation and transformation at Amsterdam, N.Y.-based St. Mary’s Healthcare. There’s also a growing need to improve communication and data exchange with other organizations.
Despite some financial pressures as a small health system, Demaree says that size can be an advantage for making workflow changes. St. Mary’s Healthcare has been “nimble” at adopting solutions to improve data access in the health information exchange, she says.
The healthcare organization adopted a new EHR platform in 2022, and AI-powered documentation tools were introduced just over a year ago. Once a patient has given consent, the tool listens, records and generate notes from the patient-doctor conversation.
The process still requires some manual work, and will always require provider review, but once St. Mary’s Healthcare implements the latest version of its EHR platform this summer, Demaree says, deeper integration will streamline workflows with additional AI functionality.
“When you build a new EHR, it's a chance to start over,” Demaree says. “As we implement the newest version of our EHR, we have the benefit of three years of experience and can build our system to meet the needs of our patients and clinicians. Of course, optimization efforts will never end. We work closely with our users to ensure that we are constantly improving.”
healthtechmagazine
