Do people and computers “think” alike?
The NY Times reports on how people and computers make a diagnosis. It turns out there’s a doctor named Dr. Gurpreet Dhaliwal, who does a kind of medical high-wire act. He comes up with accurate diagnoses on challenging cases – in real time in front of an audience. Now, not every doctor can do what Dr. Dhaliwal does – but most can do something like it.
So, what is it that doctors are doing when coming up with a diagnosis? And can a computer do it, too?
Many doctors have an “encyclopedic” knowledge of areas of medicine. But knowledge isn’t the same thing as insight – the ability to put together diverse pieces of information in a way that reveals the connections between them. A computer can be programmed with all of the medical information ever discovered, but without insight, it’s no more valuable than a book on a shelf.
Even with advances in computer “thinking,” this is where the trouble lies. Studies find that for as common as misdiagnosis may be (it’s estimated to happen between 15%-28% of the time), a lack of knowledge is rarely the cause of these misdiagnoses. Instead, the problem is usually a failure to put the pieces of the puzzle together in the right way. It’s a failure of what scientists call “synthesis,” and what the rest of us might call “insight.”
So computerized tools do have a role to play, if what they can do is trigger the doctor’s natural ability to apply insight to a problem. Tools that help doctors question their assumptions about a case would, based on the research, be highly valuable. But too much of the current focus of healthcare information technology is on prodding doctors to make sure they keep a detailed record of the fact that they, for example, counseled a patient to stop smoking. Instead, we should use the remarkable technology already available to do something far more valuable – help doctors do exactly what we really want them to do.