Maine’s largest health providers already are using artificial intelligence to streamline note-taking during visits, which they said lessens the documentation burden on providers.
But other providers in smaller settings and private practice told The Maine Monitor they are concerned about the growing role of AI in care. And at least one study had conflicting findings about how effective AI really is when helping doctors with diagnoses.
MaineHealth recently completed — and plans to expand — pilot programs with AI platform Abridge to provide ambient listening during appointments, which reduced documentation for providers, said Dr. Adrian Moran, chief medical and transformation officer.
Providers get consent from all patients prior to using the ambient documentation tool, said Dr. Daniel Nigrin, MaineHealth’s chief information officer.
Nigrin said AI helps providers by reducing rote tasks and distilling large amounts of information into summaries so that they can focus on patients instead of documentation or data entry. MaineHealth has a policy that prohibits use of publicly available AI tools when dealing with protected patient health information and has an AI advisory committee that reviews all new uses of AI technology, he said.
Moran added that MaineHealth is exploring how AI and machine learning can help reduce insurance denials.
“AI doesn’t make the final decision, but it pulls together the information we need as clinicians to make better-informed decisions on behalf of our patients,” Moran said.
Nigrin said it’s important to “move cautiously” with AI to ensure technological advancements still serve patients, but said he expects the technology to serve a “critical role” in health care in the years to come. It won’t replace doctors, said Nigrin, but providers who use it “will far outperform those who do not.”
The diagnostic capabilities of AI are improving rapidly. In a small study published last October, researchers found that ChatGPT was better at diagnosing illnesses from reading case reports than a group of 50 doctors. Doctors who used the chatbot to help them assess the reports performed only slightly better than those who did not have the chatbot.
“The study showed more than just the chatbot’s superior performance,” The New York Times reported. “It unveiled doctors’ sometimes unwavering belief in a diagnosis they made, even when a chatbot potentially suggests a better one.”
Part of the problem was doctors didn’t know how to use the chatbot to its fullest extent, The Times reported. That could change as more health care systems incorporate the technology and providers become more comfortable using it.
The providers The Monitor spoke with said they are currently using AI to streamline administrative tasks, rather than for diagnostics. Jayne Van Bramer, president & CEO of mental health provider Sweetser, said her organization is using technology from Eleos that takes notes throughout a clinical session. She called the tool a “game changer” in allowing providers to focus on client engagement and support.
“Seeing this application in action — seeing AI used like a medical scribe that follows a provider around to help with documentation demands — has reassured staff and affiliates that this is there to enhance their work,” Van Bramer said.
The application doesn’t record the session and there are no recorded clips of client engagements, according to Justin Chenette, senior director of public relations and advancement. Chenette said the tool is HIPAA compliant and providers tell clients about the device and how it will be used, turning it off if they object. AI-assisted notes are not finalized until a clinician reviews, edits and approves them, Chenette said. “AI is just helping clinicians be more present.”
Northern Light Health is using Clinical AI Agent from Oracle Health and DAX Copilot for patient visit documentation and is working on a policy for systemwide guidelines and procedures “for the ethical and responsible use of AI technologies,” said Hugh Jones, senior vice president and chief strategy and business development officer. Consent for these programs is incorporated into general patient consent, and then providers ask patients before they start using the AI scribe.
While AI can be beneficial to health care, Jones said, it will take time for the industry to fully adapt.
“One of the key barriers to adoption is the sensitivity around privacy and information security for highly sensitive [and regulated] personal health information,” he said. “These are surmountable in the long-term.”
In response to a survey on the use of AI in health care, some medical and health care providers told The Monitor they had concerns about the growing use of AI across different settings. Many are in private practice or work at smaller organizations ranging from hospital settings, in-home nursing, behavioral health and oncology.
Very few said they were currently using AI, although a couple said the technology is useful for quickly gathering studies and other research. Multiple respondents said they worried about people using AI to make medical decisions without consulting health care providers.
Sarah Kelley, an oncology social worker in Portland, said she uses AI technology on her personal computer to quickly find community groups and resources for patients across large geographic distances, but worries about AI replacing aspects of the patient-provider relationship.
“I would be very concerned if AI tried to use mental health screeners to replace sound clinical face-to-face judgement [that] I glean through relationship building and patient interaction,” said Kelley.
Elaina George, a mental health counselor, said she isn’t currently using AI and she’d prefer it wasn’t in health care at all.
“[AI] doesn’t belong in health care,” she said. “It takes the human aspect out of something very personal and unique.”
This story was originally published by The Maine Monitor, a nonprofit and nonpartisan news organization. To get regular coverage from the Monitor, sign up for a free Monitor newsletter here.
