Brian Bartholmai, M.D., chair, Radiology Informatics Division, says humans will be critical to validation and putting automated results in clinical context — “trust but verify” the output of the algorithms, and that the combination of humans and computers will do an overall better job.
Regarding a quantitative tool currently used in the Mayo Clinic thoracic radiology practice, he notes: “This tool is decision support. This is a quantitative tool that helps me do my job more confidently and more consistently.”
Ad Statistics
Times Displayed: 20219
Times Visited: 389 Stay up to date with the latest training to fix, troubleshoot, and maintain your critical care devices. GE HealthCare offers multiple training formats to empower teams and expand knowledge, saving you time and money
Humans have empathy for a patient and can explain what is going on in a way that machines cannot, he says. Plus, interventional procedures may be enhanced by machines, but will not be overtaken.
It was with this in mind that Mayo Clinic established an Office of Augmented Human Intelligence, which posits that artificial intelligence will “help people work smarter and more efficiently” to unlock “new, transformative solutions for complex medical challenges.”
For many people, technological advancements can bring uncertainty about the future — about their future, Dr. Bartholmai says. But then, as the new technologies are put to use they come to realize that there are benefits.
Artificial intelligence already guides several aspects of our lives, from curating user-specific music playlists to reading and news-consumption recommendations, he says, but once it crosses over to medicine it may start to feel a little scarier. “But really it’s the same concept.”
The natural progression from advanced and cutting-edge to normal and no big deal is evolutionary, Dr. Bartholmai says, adding the same will be true of augmented human intelligence in medicine.
“Mayo Clinic has a strong history of early clinical application of advanced technology that eventually is no longer considered cutting-edge. It’s just everyday technology now,” he says, citing the first clinical CT scanner in the US in the early 1970s, a digital radiology information system and 3rd party DICOM image archive in the early 1990s as examples.
Ultimately, both Dr. Bartholmai and Dr. Erickson say, as with other machine-assisted systems and devices, a human element will often be required.
Dr. Bartholmai says the current generation of augmented human intelligence tools Mayo Clinic and other institutions use are still generally single-purpose expert systems that lack the flexibility of the human mind — what he calls “general intelligence” — to “fuse the clinical information from the record, the intervention that we know happened, the prior CTs and the MRs and the PET/CTs and determine what should happen next for the patient.”
