‘Good friction:’ Experts share how artificial intelligence works in their office

Key takeaways:

• Replacing answering services and paper forms with AI can improve patient and provider satisfaction.
• Choosing technology that integrates with your EHR is crucial.

DENVER — Incorporating artificial intelligence into daily practice can benefit both providers and patients, according to experts here at the American Academy of Neurology annual meeting.

In a panel discussion, Louis A. Tramontozzi, III, MD, a neurologist in private practice at North Shore Neurology in Massachusetts, detailed how his team has implemented several AI-related tools to improve provider and staff productivity and wellbeing, while generating cost savings and a positive impact on patient satisfaction.

Tramontozzi’s practice, which consists of about nine providers, receives up to 800 patient phone calls per day and 20 voicemails per hour. This call volume was the impetus, he said, to firing their answering service and hiring a robot. Now, when a patient leaves a voicemail, the audio is immediately and automatically transcribed and input into the patient’s chart and assigned to a staff member.

“This has allowed us to keep the patients front and center, so our staff are interacting with patients instead of being on the phone. And they’re able to turn around some phone calls, our record is in 2 to 5 minutes, which is unbelievable,” he said. “It’s helped our staff to feel more empowered as well, and that they’re valued more [because] the work they’re actually doing is much more meaningful than listening to voicemails and transcribing.”

Panelist Leeann Garms, chief executive officer of Raleigh Neurology Associates in North Carolina, said this type of AI has the potential to create “good friction” in the system because patients and providers are interacting in a more meaningful way. For those in a position to make this switch, Garms suggested starting with night calls, then migrating to daytime calls if it makes sense for your practice.

“With any change, it’s all about messaging,” Tramontozzi said. He suggested practices make patients aware of changes — like those implemented with the voicemail software — via an email blast to portal members, or as part of the provider’s script at the end of new patient visits.

Patient intake and consent forms are another area that has been improved by AI for both Tramontozzi and Garms. One year ago, new patients in Tramontozzi’s practice were handed a clipboard, pen and various intake and consent forms, which were often only partially completed, if at all. Then, staff were tasked with prepping and inputting all the information into the electronic health record, a job they hated, he said. Now, new patients receive secure text messages with intake paperwork and consent forms to be filled out prior to their visit, and the information patients provide is automatically populated in their chart. If this system presents a challenge, the practice has iPads at the office that can be used to assist the patients.

“Our staff are no longer prepping,” Tramontozzi said. “They’re much happier and we have a higher chance of retaining them and getting them back to meaningful work, and that is really empowering for all of us no matter what job we have.”

Another bonus for Tramontozzi is that for simpler patient visits, the history of present illness, or HPI, is essentially written for him based on the information the patient provides during intake. When setting up this new system, the practice collaborated with the tech company to design intake forms for specific disease states, like headache, so the HPI would be most accurate. This allows providers to spend more meaningful time with patients, leaving more time for clarification and critical thinking, he said.

For patients who may feel frustrated with the integration of more technology, Garms said it helps to incentivize them by explaining the benefits it has on their time.

“As a selling point to the pushback of ‘I don’t want to use these things,’ that’s what the incentive is: they get their service faster, they’re spending time with the people they expected to spend time with and get what they needed at visit,” she said.

For anyone looking to implement these types of AI into their practice, Garms and Tramontozzi said the first step is connecting with your EHR vendor. They often have a list of AI vendors they integrate with. They also suggested networking with colleagues who have implemented this type of technology. But the most crucial step is choosing technology that integrates with your EHR.

“That’s the first barrier that you need to solve because you don’t want to pay extra for computer science engineers,” Tramontozzi said. “It has to be integrated.”