A panel of researchers and technology professionals agreed Wednesday the rise of artificial intelligence and machine learning was more likely to augment medical professionals’ clinical abilities rather than replace the doctors, nurses and other providers altogether.
“I think artificial intelligence, predictive modeling and all the work we’re doing should be supplementing decisions at the bedside,” said Paul Harris, a professor of biomedical informatics and biomedical engineering at Vanderbilt University. “The direction we’re heading is informed decision making that’s guided by artificial intelligence, but not necessarily taking the place of clinicians.”
Harris made the comments during a discussion at the Health:Further conference’s second day downtown. He was joined on the panel by Rob Metcalf of Concert Genetics, Damian Mingle of Intermedix and Matthew Russell of Digital Reasoning Systems.
Mingle said he believes clinicians with good personalities and bedside manners will “be the safest for the longest period of time,” as machines “will have a harder time with that.”
“More information is better,” Mingle added. “Getting that good information in the right hands of someone who knows how to synthesize that works really, really well.”
Russell said he is “pretty conservative” on what he thinks artificial intelligence and machine learning are capable of in the health care field, despite his belief that the new technologies will be “the most disruptive thing I will see in my life.”
“I think we have a long way to go before we’re completely replacing mission-critical jobs with human lives at stake,” he said. “There’s a lot of inefficiency in health care we’ll be able to squeeze out.”
Mingle offered up an example of a way doctors could use predictive analytics in treating patients. His company developed a model that could predict with accuracy above 90 percent whether a patient had developed sepsis. While they were largely successful in the endeavor, it presented its own problems.
“Ironically, the thing we had the most success with presented us with the most failure,” he said. “Experimentation is OK. The wisdom that you gain is through failed experimentation.”
Harris, who also serves as director of the Vanderbilt Office of Research Informatics, said he had encountered another type of obstacle over the years.
“The biggest problem I see out there is solving the wrong problem,” he said. “I’ve solved a number of problems that didn’t matter.”