Generative AI reduces clinical documentation time at Baptist Health

At Baptist Health South Florida, which includes 11 hospitals in Miami-Dade, Broward and Palm Beach Counties, providers were grappling with an overwhelming of managing large amounts of information emanating from patient-provider dialogues.

THE PROBLEM

One major pain point was the amount of time cardiologists had to invest in documenting patient visits. This manual documentation process was not only time-consuming but also exacerbated provider burnout and fatigue. As some of the doctors pointed out, the longer time spent on clinical documentation meant clinicians had fewer opportunities to attend to new patients.

“While commercial technologies were available on the market, they came with a hefty price tag,” said Douglas Davila-Pestana, Baptist Health’s technical manager of AI – another example of a recent trend in AI-specific job titles in healthcare

“Given the challenging financial climate many healthcare organizations, including ours, were navigating, there was an urgent need for a more and efficient system to streamline clinical documentation.

“We worked to have generative AI integrated into an AI-assisted documentation app that seamlessly blended medical transcription technology with advanced AI, specifically large ,” he continued. “This unique combination meant the AI could swiftly generate clinical notes from transcribed patient conversations.”

“Before investing, organizations should weigh the costs of available against the feasibility of developing an in-house solution, keeping in mind the unique needs and expertise available within their institutions.”

Douglas Davila-Pestana, Baptist Health South Florida

The benefit of the system was its immediacy: Clinicians could have access to comprehensive clinical notes shortly after concluding a patient visit, cutting down the lag time traditionally associated with manual documentation processes, he added.

PROPOSAL

“To counter this manual documentation dilemma, a groundbreaking proposition was made: the use of generative AI, leveraging services such as AWS HealthScribe for medical transcription and other tools like Azure OpenAI, Snowflake and DataRobot,” said Jaymin Patel, manager of data platform and engineering at Baptist Health South Florida.

“The idea was to record patient-clinician interactions with the patient’s consent, transcribe these recordings into text, and then use a large language model to generate clinical summaries in a clinical SOAP format,” he explained.

“This automation process was expected to drastically reduce the documentation time to just two to five minutes post-visit. Moreover, by integrating with Snowflake, which supports our data lake, data warehouse and layer, it was proposed that patient and appointment data could be easily pulled from our data warehouse platform to further enhance and personalize these summaries.”

GPT-4, backed by research published in the Journal of the American Medical Association, was highlighted as a promising LLM given its demonstrated proficiency in suggesting medical diagnoses. All these measures aimed to improve the patient experience, improve clinical productivity, enhance operational efficiency, and lead to significant cost savings.

“In addition,” Patel said, “the data generated by the application was planned to be stored in the Snowflake data lake layer for further analysis and application .”

MEETING THE CHALLENGE

The technology journey began with the recording of patient-clinician interactions. These recordings were sent to the AWS HealthScribe service for transcription. Once transcribed, the text was fed into the large language model, which includes GPT-4. This model, by interpreting the transcription, generated concise summaries in clinical SOAP format, which then were verified by clinicians for accuracy.

“Cardiology physicians, along with their beta tester group of physicians, were the primary users,” Davila-Pestana noted. “For the tech stack, AWS Lambda and AWS S3 were pivotal in the AI service. The moment an audio transcription file was generated, the AI Lambda Service, built on AWS, got activated.

“This automation process was expected to drastically reduce the documentation time to just two to five minutes post-visit.”

Jaymin Patel, Baptist Health South Florida

“Integration with Snowflake was essential to extract patient appointment data,” he continued. “Furthermore, once the summary was clinician-edited and approved, it would get exported and integrated with our electronic health record system. This entire process ensures timely, accurate and standardized clinical documentation.”

RESULTS

This use of AI offered significant results in three major areas, Davila-Pestana explained:

  1. Time savings. The group of physicians observe a reduction of several minutes per patient interaction, with regard to the overall time spent on clinical documentation. This newfound efficiency provides them with more bandwidth to attend to new patients, thus enhancing the patient care experience.

  2. Enhanced clinical productivity. With the AI-assisted summaries, clinicians no longer have to wait for hours as they did with previous tools. The summaries now can be generated in mere minutes, resulting in quicker turnaround times and more streamlined operations.

  3. Cost efficiency. By adopting this approach, Baptist Health South Florida sidestepped the considerable expenses of commercially available options, tapping into the organization’s intellectual resources and technology and capitalizing on the efficiencies of AI.

ADVICE FOR OTHERS

“For healthcare providers contemplating the incorporation of generative AI technology, it’s crucial to view AI as a tool that augments, rather than replaces, human input in clinical documentation,” Davila-Pestana advised. “AI can bring about tremendous efficiencies, but blind reliance can lead to oversight.

“Ensuring a human-in-the-loop verification process, where clinicians review and validate the accuracy of AI-generated summaries, is paramount to maintain the integrity of patient records,” he added. “Furthermore, before investing, organizations should weigh the costs of available vendors against the feasibility of developing an in-house solution, keeping in mind the unique needs and expertise available within their institutions.”

Lastly, any AI implementation should prioritize patient consent and data security to maintain and adhere to healthcare regulations, he said.

Follow Bill’s HIT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.

Source link