By Irene Yeh and Allison Proffitt
April 16, 2024 | At the Precision Med TriConference 2024, Illumina Ventures partner Mara Aspinall sat with Eric Topol, endowed chair and executive VP and professor at Scripps Translational Science Institute, in a fireside chat to discuss precision medicine and artificial intelligence (AI). With three published books and more than 1,200 peer reviewed publications, Topol is one of the industry’s leading minds on digital medicine, and from his perspective, AI needs to be adapted sooner. But is the medical community ready for that?
“I think it's fair to say that this is going to be the biggest transformation in the history of medicine, but we're not there yet,” commented Topol. “It’s still in the very early stages because we’re missing compelling evidence in real-world medicine. Until we get that, we have to be circumspect about the progress that’s being made. But the proof points are coming pretty fast.”
The Rapid Breakthroughs of AI
AI—especially generative AI—has made great strides since it hit the scene—and it’s not showing any signs of stopping. AI models can now aid with precision medicine. By taking all the layers of a human being’s unique data, physicians and researchers can now view biomarkers to determine signs and symptoms of various diseases. It can even help with “organ clocks.” Using AI to assess over 10,000 plasma proteins, physicians can track the age of the brain, heart, kidneys, liver, and more. Instead of putting a patient through multiple scans, medical professionals can just use markers, blood tests, and multimodal AI to conduct these readings and determine if a patient is at high risk for certain conditions, thus allowing earlier intervention.
Another breakthrough of AI is the quality of the natural language processing aspect. AI can now generate a high-quality synthetic note of a patient’s exam. With the patient’s consent to have the exam recorded, the AI provides the patient with an audio file and the notes so they can review what occurred during the visit and keep a record for themselves.
But from Topol’s perspective, perhaps the most “unexpected” breakthrough is how AI has helped physicians become more mindful and empathetic of their patients. “These notes… will tell the doctor, ‘Why did you interrupt Ms. Jones after nine seconds?’ If you ask for coaching, they will tell you that you didn’t show any sympathy for this or that. You didn’t let this patient express their concern. And it’s promoting empathy like we’ve not seen before.”
Topol expects that in the next few years, every clinician will have to go through coaching with the AI to improve communications and empathy with patients. Though some physicians may be hesitant to take on the coaching, Topol predicted that such training will be similar to a Compulsory Medical Examination. He also encouraged reluctant physicians to look at it as a “rescue mission” toward “keyboard liberation.” Physicians spend “three hours of data clerk function a day,” which deprives them of interacting with their patients face-to-face. Such a tedious task can also lead to burnout, which prevents medical professionals from properly helping patients.
“We don’t need brainiacs in the future,” said Topol. “We need people who have really great interpersonal skills, who want to be coached better.”
Another surprise was how multiple groups of people felt more comfortable sharing their personal information and issues with an AI than with a human being. This can be used to the medical field’s advantage to get a better understanding of the patient’s health conditions and state of mind. It also has the potential to reduce biases around cultural background and gender, which is greatly beneficial for underrepresented groups.
Despite these “proof points” of AI’s rapid development, adapting AI in the medical field has been slow and even met with resistance. Medical professionals are reluctant to adapt to the changes. For example, even though telemedicine was a tool that could have been commonly used earlier, it took the COVID-19 pandemic to make it widespread.
This reluctance also extends to medical schools. Currently, none of the 150 medical schools in the U.S. have AI on the curriculum. “The medical school curriculum hasn’t changed for a long, long time,” Topol said. He and Aspinall suggested adding AI, as well as personalized medicine and genomics, onto board exams so that medical professionals are forced to keep up to date with current information.
“I don’t know that we need a crisis to affect the change,” said Topol. “I think all we need is, you know, the extraordinary evidence to say this is the way practice should be altered. And according to that, the handwriting is on the wall.”
Topol is not a lone voice, though. Other physicians at the event were sharing their own efforts to incorporate AI into their practice. In another plenary fireside chat, Kevin Davies, executive editor of The CRISPR Journal, interviewed Doug Flora, executive director of St. Elizabeth Healthcare and editor of a new journal, AI in Precision Oncology, shared his experiences converting his colleagues—and patient populations—to the impact of AI. Flora is a practicing oncologist serving rural communities in Ohio and Kentucky, drawing patient populations mostly from Appalachia.
It’s a challenging spot for a physician, Flora said. “It’s a dark place for healthcare,” he said, frankly. “There’s very poor health literacy... There’s a very high degree of obesity. Lots of alcohol. Lots of tobacco. About 29% of people still smoke there. We’re center of the universe for lung cancer. About 25-30% of our patients are small cell lung cancer, which is smoking driven.”
But even in this challenging landscape—far from big urban medical centers—Flora has found AI to be a powerful tool.
Flora mentioned he was actually introduced to AI through Topol’s books, and reading Topol’s book Deep Medicine set off a period of personal exploration. “I’ve taken about every available course; I’ve read everything I can... It’s not as daunting as it might seem from the outside,” he said. “Clinicians have been suspicious, but ultimately worn down and responsive because they need these tools as much as the rest of us.”
Perhaps unsurprisingly, Flora’s view of the future of AI is quite aligned with Topol’s. “It’s a tool like any other tool. You wouldn’t expect your CFO of a hospital system or major multinational to work without Excel. We can’t function without some sort of slide deck anymore. This is another tool. It’s not going to solve all of our problems, but it might accelerate the solving of quite a few of them.”
Many opportunities for physicians to use AI are not specifically in the delivery of care. For instance, Flora says generative AI saves him 2-3 hours each day, “using either GPT or one of the new cloud models, everything I write—every email I answer, every time I fight with an insurance company to approve a PET scan or a therapy that I thought was necessary for my patient. It used to be 40 minutes of me typing—because I type very poorly—but now I have prompts for everything.”
Flora acknowledged that many doctors are deeply distrustful of AI. He recounted recently being part of the American Society of Clinical Oncologists’ effort to build a strategic plan around AI. “There were some brilliant, brilliant providers, world-class physicians, kind of giants in my field and they were really dug in about little things like content generation for social media. And it was hard to predict... They were just railing against the machine,” he recalled.
Topol has seen the same responses. “I think the biggest mistake I’ve made over the years is expecting it to happen a lot quicker than it ever does. That, I think, is a real problem because the medical community is kind of sclerotic, ultra conservative, and resistant to even not just change, but oftentimes education of new fields. And it’s really a challenge for all of us.”
But there is good news. Many ASCO doctors are on board, Flora said, and the strategic plan is moving forward, reiterating that progress is happening.
There’s an opportunity here not to miss, Flora warned. Adopting new technology early means clinicians get a voice in how best to use it. “We, as providers..., let the electronic medical record be taken away from clinicians and it became billing software,” he said. “I don’t want that to happen to AI.”