Pierre Elias starts most of his presentations about clinical informatics with a child’s drawing: It shows a doctor typing on his computer with his back to his patients, while a child holds a doll in the background and her mom is seated on the exam table.
To Elias, it encapsulates everything that’s wrong about medicine’s relationship with computers.
“What happens when a generation of people grow up with this as their expectation of a doctor’s visit?” he asks. “We deserve better technology in health care.”
Elias said he is appalled by medicine’s continued reliance on the fax machine. It is an angst made exponentially worse by time he spent on the other side of the digital divide in Silicon Valley, where he worked for a startup whose technology was acquired by Google.
He ultimately decided to go back to medical school, where he could apply his informatics knowledge to a field where it could save lives. Elias is now a cardiology fellow at Columbia University’s Irving Medical Center, where he is researching the use of machine learning to improve care for cardiac patients.
“An electrocardiogram alone has 60,000 data points,” he said. “There are going to be pieces of information that lay within those pixels that we as human beings have not understood are important.”
Finding patterns within those data could allow doctors to get earlier warnings of heart function problems that often lead to heart failure and death. As much as he believes in the power of the technology, Elias also believes it must be rigorously evaluated before it is implemented in clinical practice.
“What happens when a generation of people grow up with this as their expectation of a doctor’s visit? We deserve better technology in health care.”
“We have to treat these like digital therapeutics,” he said. “We have to put them to task and ask, how good is this stuff? Can you prospectively validate that these deep learning algorithms can find people who need our help?”
— Casey Ross
