In the next installment of our Medscape Masters event series, we bring together a unique group of experts to discuss how artificial intelligence (AI) is changing the landscape of medicine, moderated by John Whyte, MD, WebMD's chief medical officer. Joining our panel is David Fajgenbaum, MD, a physician, scientist, author, and associate professor of medicine at the University of Pennsylvania's Perelman School of Medicine; Elaine Mardis, PhD, co-executive director of the Institute for Genomic Medicine at Nationwide Children's Hospital; and Leo Celi, MD, a principal research scientist at MIT and an associate professor at Harvard Medical School.
Fajgenbaum is pioneering the use of AI to help repurpose drugs designed to treat one disease and expanding their potential applications to treat a wide range of other diseases. Using AI, Fajgenbaum and his team can analyze thousands of drugs and assess their potential to treat over 18,000 human diseases -- an endeavor that would be impossible for any human team to complete. Remarkably, AI can achieve this in just 24 hours.
Fajgenbaum, who was diagnosed with Castleman disease -- a rare disorder causing noncancerous tumors in lymph nodes -- during medical school, applied AI to identify a new treatment option. Through this AI model, he found that a tumor necrosis factor (TNF) inhibitor that had never before been used to treat Castleman's was the key to saving one patient's life when all other treatments failed.
It's still crucial for human beings to be involved in these analyses, said Fajgenbaum, especially when it comes to assessing for drug-drug interactions.
Mardis feels that AI algorithms devoted to cancer diagnosis and treatment are driving what she calls "precision medicine." This works really well when AI is involved in evaluating the genomics of cancer patients coming out of clinical trials.
"What are the other factors in that cancer genome that are helping to predict response, durability of response, and, importantly, the potential for combination therapies?" asked Mardis.
According to Mardis, when adult patients are diagnosed with cancer by their doctor, in many cases they should also get a second opinion that is driven by AI models -- even more so for patients diagnosed with a very prevalent and frequently diagnosed cancer type.
Celi highlights that the results of a clinical trial conducted in a single population cannot be assumed to apply universally. However, AI models enable the continuous development of trials tailored to specific populations, overcoming this limitation.
"It's very expensive to repeat a clinical trial that would engage all types of patients across all types of diseases," said Celi. "It's just economically and logistically impossible."
But with the growing number of electronic health records now being amassed, he said, we can use that information for clinical trials going forward. The issue that remains, though, is how to avoid AI-driven biases that might skew the data to reflect only a certain group. The key to overcoming this is transparency -- understanding how the data were collected, who they were collected from, and what equipment was used to collect them.
"The clinicians have to be at the front and center," Celi said. "We will only have a successful deployment of AI if we are taking the reins of AI and taking them away from the tech giants."
This year's Health Heroes are using AI to find cures, prevent deadly illnesses, and flag cancers before they start -- and for them, it's personal.
