Rise of the machines

Gawande takes it one step further -- if expertise means mind-numbing routine, can human judgment be replaced by computers? He cites a big EKG study pitting a world-renowned cardiologist vs. Lars Edenbrant's diagnostic algorithms (a pity Lars didn't call it Deep Red or ENIAKG or something).

The computer caught more abnormal EKG's than the cardiologist. And when it comes to things like appendicitis, where surgeons are expected to operate unnecessarily 15% of the time, computers can probably do better, too. Humans are distracted by things like "that memorable case last week" and "the patient's mom was really scared" ... computers know just the right weight to put on disparate data points like age and WBC and fever and vomit.

This leads to two scenarios for doctors: either factory-style expertise in one very narrow field like hernia repair, or acting as a liaison between the patient and the computer diagnostician.

On a second look, the liaison-scenario isn't so bleak, or so different from what goes on now. In family practice, a lot of visits are devoted to collecting information, going over test results, educating patients about different options, and counseling on health problems. In the future, we'll simply be adding computer diagnoses into the mix (my palm already has the Framingham Heart MI calculator -- just plug in cholesterol and age and blood pressure and other variables, then find the risk of heart attack over the next few years).

In fact, Gawande suggests doctors will have more time with patients this way -- more time to interact and explain things, less time spent coding bills and writing up differentials. I'm not sure that's true, but in terms of time spent with patients, it beats the focused-factory approach.

The bottom line is improved outcomes and patient ("customer") satisfaction. Who wouldn't go for more time with primary care docs, less time waiting for surgery, improved diagnoses, and reduced complications?

Maybe doctors will oppose it. I think my fellow med students and I still share a bit of the romantic notion of the generalist physician -- seeing all kinds of patients, diagnosing everything from papilledema to ankle fractures. If the idea of narrow factory-style specialization, or turning over diagnosis to the computer, is unsettling for us, imagine how tough it is for doctors already practicing.