Topol: AI-fueled ‘keyboard liberation’ will improve the medical encounter

Key takeaways

• AI brings the promise of medical error reduction and improved diagnostics.
• AI may restore the patient-doctor relationship.

BOSTON — “We should be embracing [AI] and trying to make this a priority to get this right, because I don’t know any alternative to get us out of this frustration right now outside of this technology,” Eric Topol, MD, said.

“The opportunity is vast ... and we must eke out the advantages,” Topol, who is the Founder and Director of the Scripps Research Translational Institute, said during the ACP Internal Medicine Meeting keynote.

AI has already shown promise in medicine, particularly in specialties with advanced diagnostics and screening.

“Every type of medical scan can be improved by [artificial intelligence (AI)],” he said.

Topol said the Johns Hopkins Group has estimated 800,000 people die or are permanently disabled per year as a result of medical errors.

“The name precision medicine is very popular [but] ... we need accuracy in medicine” and that is where AI can be harnessed. Topol listed using AI for mammography readings, colonoscopy machine vision and retina imaging to diagnosis nonophthalmic diseases, such as diabetes, BP control, kidney disease, liver and gallbladder disease, heart calcium score, Alzheimer’s disease, heart attack, stroke, hyperlipidemia and Parkinson disease.

He said the retina is the “gateway” to diagnosis.

Additionally, Topol discussed AI’s potential in the future such as “doctorless diagnosis”; coaching for chronic disease; and “rebooting cancer screening with AI” by evaluating multilayer risk and weighing it against the need for screening.

Topol said AI can be used to restore the patient-doctor relationship because the electronic health record administrative burden has made medicine “fast and shallow.”

Using AI for ambient documentation can provide “keyboard liberation” and give doctors more opportunities for eye contact and “laying of hands” during the medical encounter, which can also benefit the mental health of the physician, he said.

Topol said there are studies coming out that demonstrate AI can be used to create better management plans and have more empathy toward the patient.

“I thought we would never say that a machine would have more empathy,” he said.

But the completeness of the differential diagnosis, clinical reasoning and directness of information provided with AI tools can be better than the doctor with one major caveat, according to Topol.

The major caveat is that all this research is contrived because it uses retrospective data, he said.

“This is not real world,” Topol said, which led him to warn the audience about the need for regulation and their own responsibility when using AI tools.

“When are you going to give us some teeth, FDA?” Topol asked as it relates to the regulatory environment, oversight and the need for trusted evidence of AI technology.

“I hope the FDA gets more stringent because these studies don’t get published, there is no transparency,” he said noting that if there is no trust in the data, “we’ve got a recipe for trouble.”

The audience clapped at this comment.

Oversight and transparency are key because the physician is ultimately responsible for the use of AI and its outcomes, he said.

And, yet Topol encouraged physicians to lean into AI because there is a lot of good that can come from its use, including the opportunity to reduce burnout, and called the audience to “steer it, manage it.”

“We have the means to get us to caring for patients again,” Topol said.