sourcegraph
April 13, 2024

Dr. Matthew Hitchcock, a family physician in Chattanooga, Tennessee, has an artificial intelligence assistant.

It records patient visits on his smartphone and aggregates this information for treatment planning and billing. He made some light edits to the AI-generated content and completed his daily patient visit notes in about 20 minutes.

Dr. Hitchcock used to spend up to two hours printing these medical records after the four children fell asleep. “That’s a thing of the past,” he said. “It’s amazing.”

The grand vision that ChatGPT-style AI is entering healthcare is inspiring. Enthusiasts predict that every doctor will have a super-intelligent assistant offering suggestions to improve care.

But first will be the more mundane applications of artificial intelligence. The main goal is to relieve the heavy burden of digitized paperwork that doctors must produce, entering lengthy notes into electronic medical records needed for treatment, billing and administration.

For now, the new artificial intelligence in healthcare will be less of a genius partner and more of a tireless scribe.

From the leaders of major medical centers to family doctors, there is optimism that healthcare will benefit from the latest advances in generative artificial intelligence — a technology that can generate everything from poetry to computer programs, often with human-level fluency.

But doctors stress that medicine is not a vast field of experimentation. It might be fun for AI to occasionally create disinformation or supposed hallucinations, but not in the high-stakes world of healthcare.

This, they say, makes generative AI very different from the AI ​​algorithms already approved by the Food and Drug Administration for specific applications, such as scanning medical images for cell clusters or subtle patterns that indicate the presence of lung or breast cancer. Doctors are also using chatbots to communicate more effectively with some patients.

Regulatory uncertainty and concerns about patient safety and litigation will slow the uptake of generative artificial intelligence in health care, particularly its use in diagnosis and treatment planning, doctors and medical researchers say.

Doctors who have tried the new technology report dramatic improvements in performance over the last year.Medical records software designed so doctors can examine AI-generated records Based on what the patient said during their visit, make it verifiable and foster trust.

“At this stage, we have to choose use cases carefully,” said Dr. John Halamka, president of platforms at the Mayo Clinic, who oversees the health system’s adoption of AI. “Reducing the documentation burden is a huge win in itself.”

Recent research shows that doctors and nurses report high burnout, hint Many people have left the industry. For primary care physicians in particular, topping the list of complaints is time spent on electronic health record documentation. This work often extends into the evening, the hard work after get off work, which doctors call “pajama time.”

Generative artificial intelligence appears to be a promising weapon in the fight against the doctor’s workload crisis, experts say.

“At a time when healthcare needs help, this technology is improving rapidly,” said Dr. Adam Landman, chief information officer at Massachusetts General Hospital, which includes Massachusetts General and Brigham and Women’s Hospital in Boston.

Physicians have used a variety of documentation aids for years, including speech recognition software and human transcribers. But the latest AI is doing much more than that: Summarizing, organizing and labeling conversations between doctors and patients.

Companies developing such technologies include abridged, Atmosphere Healthcare, Ogg Madix, nuancewhich is part of Microsoft, and Suki.

Dr. Gregory Ator, an ear, nose and throat specialist and chief medical informatics officer at the University of Kansas Medical Center, said 10 doctors at the University of Kansas Medical Center have been using generative artificial intelligence software for the past two months. The medical center plans to eventually provide the software to 2,200 physicians.

But the Kansas health system is avoiding using generative AI in diagnostics, fearing its recommendations could be unreliable and reasoning opaque. “In medicine, we don’t tolerate hallucinations,” Dr. Atto said. “And we don’t like black boxes.”

The University of Pittsburgh Medical Center has been a test bed for Abridge, a startup led and co-founded by Dr. Shivdev Rao, a practicing cardiologist and executive in the medical center’s venture capital arm.

Abridge was founded in 2018 when large-scale language models (technical engines that generate artificial intelligence) came into being. Dr. Rao said the technology opened the door for automated solutions to paperwork overload in healthcare, which he saw around him, even with his own father.

“My father retired early,” Dr. Rao said. “He just couldn’t type fast enough.”

Today, more than 1,000 physicians at the University of Pittsburgh Health System use Abridge software.

Dr. Michelle Thompson, a family physician in Hermitage, Pa., who specializes in lifestyle and integrated care, said the software has freed her up to nearly two hours a day. Now, she has time to go to a yoga class, or sit down to a family dinner.

Another benefit, Dr. Thompson said, is an improved experience for patients visiting the clinic. No more typing, note-taking, or other distractions. She simply asks patients to allow her phone to record their conversations.

“As a doctor, AI allows me to serve patients 100 percent of the time,” she said.

AI tools also help patients become more involved in their own care, Dr. Thompson added. Immediately after their visit, patients receive a summary that can be accessed through the University of Pittsburgh Health System’s online portal.

The software can translate any medical term into plain English with a reading level of about fourth grade. It also provides a record of visits and color-codes “medical moments” for medications, procedures, and diagnoses. Patients can click on colored labels and listen to parts of the conversation.

research shows The patient forgetting rate is as high as 80% What doctors and nurses say during visits. Dr. Thompson said the recorded AI-generated visit summary is a resource her patients can return to for reminders to take medication, exercise or schedule a follow-up visit.

After the appointment, the physician receives a summary of the clinical notes for review. There are links back to records of doctor-patient conversations, so the AI’s work can be checked and verified. “It really helped me build trust in AI,” Dr. Thompson said.

In Tennessee, Dr. Hitchcock, who also uses Abridge software, read reports of ChatGPT scoring high on standard medical tests and heard predictions that digital doctors would improve care and address staffing shortages.

Dr. Hitchcock tried ChatGPT and was impressed. But for legal, regulatory and practical reasons, it would never have occurred to him to load patient records into a chatbot and ask for a diagnosis. Now, he’s thankful that he’s free at night and not bogged down in the tedious digital documentation that America’s healthcare industry demands.

He doesn’t think any technology will solve the health care shortage. “AI is not going to solve this problem anytime soon,” said Dr. Hitchcock, who is looking for another doctor to provide care to his four doctors.



Source link

Leave a Reply

Your email address will not be published. Required fields are marked *