Healthcare AI integration is a way of forcing medical schools to rethink training physicians and shifting from memory of fact to developing skills that complement AI technology.
Artificial intelligence is causing several dramatic changes to healthcare. Whether it helps doctors analyze medical images more quickly, suggest more personalized treatment options, or help biopharmaceutical companies develop new treatments, AI continues to entrench itself in all aspects of the industry. This was one of Hartford Healthcare's first takeaways from August's AI summit. Another is that the way doctors and medical technicians are trained must also be changed to keep up with the rapidly evolving use of the technology. That point was raised by Dr. Nina Dutta, chairman of Hartford Hospital's medical school during the summit. “AI is now integrated into all levels of our work,” says Dutta. “I think that's amazing, but it makes me think – what should we do in terms of training future doctors? What skill set do they need?” She said the medical training she and others have just received seemed a little outdated. “We were really taught that we can put a lot of knowledge in our brains and spit it out, and the people who can spit most of the information were the best doctors,” Dutta said. “I don't think that's true anymore, but I'm not sure where to focus on this future.”
The two summit panelists came from companies that provide AI-related systems to the healthcare industry. We're Ran Shaul, co-founder and chief product officer of K Health, and Elad Walach, AIDOC CEO. They each had suggestions to improve medical training. Wallach suggests that doctors need to learn statistics, while Shoal argues that providers need to learn about algorithms. “That would be a must,” Shoal said. “In my opinion, we need to understand the sensitivity of different data types and different models of working models. …That's how we embedded it into the thought process. There's no shortcuts there.” Dr. Padmanavan Premkumar, Senior Vice President of Hartford Healthcare and president of HHC Medical Group, said he warned that doctors were already trying to deal with the overwhelming amount of information about patients. “There's too much information to help our clinicians function in front of patients today,” he said. “The backend has to find ways to aggregate data that makes sense to us.” Given the challenges presented to both doctors, medical schools and healthcare facilities, we are currently looking at ways to integrate AI into training.
“An active change of life”
Dr. Peter Yu, Chief Academic Officer at Hartford Healthcare, says it's clear that AI will continue to change the way physicians work.
“It has been said many times that doctors don't expect to be replaced by AI,” Yu said in a recent interview. “But doctors will be replaced by doctors who know how to use AI.” Over the past 20 years, medical training has shifted from “every time you need to carry all the knowledge of medicine in your mind,” he said, to having general knowledge of available information. “Our youngest colleagues today don't always rely on their memories to remember information,” Yu said. “All they need to do is understand how to access good information and make good decisions based on it,” he said in the past, the development of the internet, searchable databases and online journals has dramatically changed healthcare and education. The transformations currently occurring include techniques that do some of the information gathering and analysis that people have previously performed.
“New colleague”
Dr. Barry Stein, Vice President and Chief Clinical Innovation Officer at HHC, said he has established the Healthcare Innovation Center for AI Innovation. First, we can combine young clinicians and academic partners with “influenced leaders” from healthcare organizations to “attack on issues that are important to clinicians,” so we can “see ways that technology can help solve problems that cannot be solved,” he said. Second, even in a “safe and organized environment,” Stein said, it shows doctors that technology could be wrong.
For the third aspect, Stein cited the concepts raised in the book “Co-Intelligence,” written by Ethan Morrick of the Wharton Business School at the University of Pennsylvania. In the book, Mollick suggests viewing AI as “the new colleague I've met for the first time,” Stein said. “Essentially, what we say to clinicians is treating this technology like new residents like new students. We don't know what they know. We don't know how dangerous they are. Always use critical thinking to evaluate this new colleague.” With these concepts being introduced, Hartford Healthcare is ready to help medical schools weave AI training into their curriculum. This is a role that Frank H. Netter MD School of Medicine at Quinnipiac University in North Haven was recently invited to undertake.
Please proceed with caution
Dr. Lisa Coprit, interim dean of Kinnipiac's medical school, said she recently met with Stein to discuss creating the necessary curriculum that integrates AI into medical education and patient care. Coplit said AI is already transforming healthcare and education. Generated AI helps clinicians streamline workflows, expand access to care and “enrich student learning,” she said. “The next frontier, Agent AI, could infer independently,” Coplit said. But the healthcare industry also needs to approach AI with caution, she said. “The issues of governance, safety, errors and cognitive desk risk are realistic and require thoughtful surveillance,” Coplit said. “That is why AI education must be closely related to the human dimension of medicine with a new emphasis. No matter how advanced technology is, the mind of medicine is human connection.”
Upgrade time
Dimitris Bertsimas, vice-president and vice-dean of business analytics at Massachusetts Institute of Technology, has consulted with healthcare organizations for 30 years, including Hartford Healthcare for the past 10 years. Medical training has not changed since the 1920s and has been delayed for a long time due to upgrades, according to Bertsimas. This should include teaching future physicians the advantages and disadvantages of AI. He noted that AI has made great strides by reading mammograms, for example, “with overall higher accuracy than humans.” AI can also help doctors make a diagnosis, and improvements are being made with AI-reading ECGs, he said. “We have the full range of proper AI curriculum and there's more going on in a few minutes,” says Bertsimas. “I think the year's knowledge base will be much more than this year, but in two years it will be even more and more and more abilities.”
