Ochsner Health turns to AI to help clinicians reply to patient messages

Louisiana’s largest not-for-profit health system began piloting a new generative artificial intelligence (AI) tool in September to help providers more efficiently respond to patient questions.

About 100 clinicians across Ochsner Health’s 46 hospitals and 370 centers have started using technology from Microsoft that integrates electronic health record (EHR) data to draft and send “simple messages” to patients.

“[The AI] will reduce [the] time our clinicians are spending on computers so that they can spend more time doing what they do best—direct patient care,” Ochsner’s Chief Application Officer Amy Trainor said in a statement.

How it works

Microsoft’s Azure OpenAI Service, which works together with the EHR system Epic, can generate responses to routine questions sent through the MyOchsner app that aren’t related to either clinical judgment or diagnoses, according to the health system.

Clinicians then review the AI-generated messages for accuracy before sending the response to patients. Ochsner emphasized that the AI tool is intended to support—not replace—communication between patients and providers.

“Our online portal, MyOchsner, allows our patients and clinicians to communicate in a very special way,” Trainor told Healthcare Brew in an email. “We want to continue that important relationship while using AI to foster it, making the patient happier with a timely answer, and making the provider happier with an easier way to get the message completed.”

The technology is HIPAA-compliant, and messages are encrypted to ensure patient privacy, according to the health system.

Message overload

Last year, the system’s physicians received more than 4 million requests for medical advice via the MyOchsner app, according to the health system. That administrative burden, according to the American Medical Association, can lead to burnout.

Providers who got more than 307 messages per full-time workweek—the highest amount the researchers analyzed—were about 6x more likely to experience exhaustion compared to those who received fewer than 147 messages, a 2020 study of primary care practices in a large health system in San Francisco found.

Health systems are also incorporating AI in other ways to reduce administrative tasks. For example, San Francisco-based Carbon Health integrated an AI-enabled notes assistant into its EHRs to help clinicians save time filling out medical charts, Healthcare Brew previously reported.

Ochsner will test the AI messaging feature over three phases this fall, adding new providers and staff, and collecting patient feedback during each phase to make changes to the technology as needed, Trainor said.

“AI holds great promise for innovation in healthcare, and we are already imagining new ways we can use AI to improve the health of our communities,” Denise Basow, Ochsner’s chief digital officer, said in a statement.