Next year, when IBM's Watson computer competes live against human contestants on Jeopardy, think of this: The machine might be your next doctor.
A projected doctor shortage has been making headlines in the U.S. since the passage in March of health-care reform, which is expected to give 32 million more Americans health insurance by 2014. At the same time, enrollment in medical schools is going down. One report by the Association of American Medical Colleges predicts a shortfall of 150,000 doctors in the U.S. in the next 15 years. Europe, Japan, New Zealand and other countries are facing similar challenges in health care.
Most of the proposed solutions involve incentives to get more students into medical schools. There's another way to attack the problem: create artificial intelligence doctors that can handle a lot of routine medical questions, easing the traffic to doctors to check out a rash or a child's fever. Sure, robots may not have the same bedside manner as human doctors. But the idea is that they wouldn't replace doctors, but would allow the human physicians to focus on more serious care.
Building a Robot Doctor
The idea is not so far-fetched. Technology is emerging that could handle some medical diagnosis through a laptop or smart phone. And there's no question that a robot computer could be loaded chock full of information. IBM's massive Watson computing system, for example, has been loaded with millions of pages of information from everything from classic novels to Wikipedia to books of historical sports statistics.
But being a doctor requires more than just knowledge. At the heart of Watson is software called DeepQA, which allows the computer to understand Jeopardy's often-tricky questions, and in a split second find the most likely answer out of millions of possible answers in all that stored data. The software will be tested in public sometime in 2011, when IBM's Watson is expected to play on Jeopardy. As it is, the massive computing system has been playing practice rounds in a mock studio against former Jeopardy winners, and Watson has been winning more than half the time.
IBM believes DeepQA will have plenty of applications beyond talk shows – including health care. Imagine if the same kind of system was loaded with every medical reference book and scientific study and information about prescription drugs, along with the latest news about flu outbreaks, weather or other factors that might impact health. A patient could then access a DeepQA doctor over the Web, ask questions and get informed and knowledgeable answers.
There's already a crude version of this on the Web – a site called Doctor Robot, created by Russian physician Aleksandr Kavokin. It asks patients to fill out a questionnaire, and attempts to match answers with a possible diagnosis.
Robot Doc's Special Features
Other new technologies could also help create a sophisticated computer doctor. Scientists keep improving machine vision, which enables a computer to analyze an image – so you could send a computer doctor an image of your rash, for instance. Wearable devices like Fitbit can track exercise and calories burned and automatically send the data into the network for use by a computer doc. Apps on smart phones could help patients keep track of other health information and send it into the system.
And efforts continue to make individual health records electronic, so a computer doc would have access to your medical history. Companies such as 23andMe offer genetic analysis that can identify predisposition for certain diseases, something else that could help an online physician with a diagnosis.
The X Prize Foundation is doing its part to make sure computer-based doctors become reality. The X Prize, which funded the contest to build an aircraft that could reach space and return to land at an air strip, this summer announced a $10 million prize to create an artificial intelligence physician. The prize will go to "the first team to build an artificial intelligence system that can offer a medical diagnosis as good as or better than a diagnosis from a group of 10 board-certified doctors." It also should be accessible through a cell phone.
Of course, there are major concerns about privacy and the acceptance of such systems by the medical profession and general public. The big question will be whether society is ready to trust health care and health data to the network. But the doctor shortage may force those issues. Long waits at doctors' offices and primary physicians refusing to add new patients might convince the public to try computer doctors, especially if Watson does well on his test.
Will Your Next Doctor Be a Robot?