When a clogged artery landed Peter Szolovits in the hospital for a coronary bypass operation in mid-October, he noticed a few incongruities other patients might not have. Machines that performed intertwined functions—dosing and delivering medication, for example—did not communicate with one another, and patient statistics detailed on paper were not in the hospital's electronic medical records.

As head of the Massachusetts Institute of Technology's Clinical Decision Making Group, which works to apply artificial intelligence (AI) to medicine, Szolovits knew that intelligent systems could optimize care by working together better to eliminate errors as well as avoid repetition of medical tests. Indeed, in the midst of the U.S. health care debate, some experts say that AI could lift some of the burden on physicians by helping them diagnose conditions and choose treatments.

Of course, the same claim echoed in the 1970s and 1980s, when a media blitz put medical AI on the cover of newsweeklies. Although it made inroads via various diagnostic programs such as INTERNIST and MYCIN, intelligent technologies did not revolutionize clinical care by saving lives, money and time.

Hurt by hype
One major problem was unrealistic expectations, remarks Edward Shortliffe, president of the American Medical Informatics Association. Integrating separate electronic medical records, for example, is complicated because the two sources may not share terminology and language. Usability was an issue, too: Early programs that helped physicians make diagnoses were inconveniently located outside patient rooms.

Today's AI researchers have taken such criticisms to heart and developed more appropriate software. One program that helps doctors make more accurate diagnoses was recently tested in a study conducted by investigators from the Mayo Clinic in Rochester, Minn. They entered lab test results and vital signs from 189 patients to train and test a program to assess whether subjects had a heart infection known as endocarditis. The infection and its complications kill 60 percent of the 29,000 people who develop it in the U.S. annually; tests for the condition are invasive, and can be painful and dangerous. But the software was able to definitively determine that half of the patients did not have the infection—eliminating the need for an unnecessary and risky procedure.

Real diagnoses with artificial networks
The software was based on an artificial neural network, a program that mimics the structure of biological brains and learns via adjustments in the strength of connections in its network. Researchers taught the software to recognize endocarditis by using information from medical records of patients once suspected of having the condition. The network learned to correlate each patient's unique symptoms with a diagnosis. "The network recognizes patterns," says M. Rizwan Sohail, an infectious disease expert at Mayo Clinic and lead author of the study, presented at the Interscience Conference on Antimicrobial Agents and Chemotherapy in San Francisco in September. "Just like humans, once we see the disease on a person a couple of times, we tend to associate symptoms with certain diseases."

A similar program to diagnose diseases that are more prevalent than endocarditis, such as pneumonia, would provide the greatest cost savings, and Sohail says that is one of the logical next steps in this research. "Diagnoses that are more common would show the value of the network by saving money and being helpful for the public and physicians," he says.

7 Comments

1. 1. dbtinc 04:12 PM 1/12/10

   The sooner we get something in AI for Dx the better most patients will be.

   Reply | Report Abuse | Link to this

2. 2. Archimedes 04:39 PM 1/12/10

   Hippocrates, the ancient Greek father of medicine, was correct when he stated:
   "LIFE is short, and Art long; the crisis fleeting; experience perilous, and decision difficult. The physician must not only be prepared to do what is right himself, but also to make the patient, the attendants, and externals cooperate. Hippocrates"
   Of course, Hippocrate's profound statement implies that artificial intelligence would necessarily be an extremely important and efficacious tool for physicians given the "perilous experience and difficult decisions" inherent in medical reasoning.

   Reply | Report Abuse | Link to this

3. 3. Spoonman 12:40 PM 1/13/10

   The problems presented in the article are all surmountable. The problem is, it appears everyone's paying attention to the long game without looking at what low-hanging fruit can be grabbed now. For example"
   
   " Machines that performed intertwined functions—dosing and delivering medication, for example—did not communicate with one another, and patient statistics detailed on paper were not in the hospital's electronic medical records."
   
   Mentioned elsewhere in the article, it states that many of these issues are due to differing "languages" being spoken by each machine. No problem, the first step to solving the over arching problem is to fix the communication issues. Come up with a protocol for inter-machine communication and networking and start hooking the machines together. There doesn't need to be any AI at this point, so the technology and ability to do this is not only here, it's been in use for ages.
   
   Similarly, coming up with a system that allows doctors to more easily utilize electronic records will result in them being used more consistently and effectively. But, again, it needs to follow a universal protocol that ALL electronic records vendors need to follow for interoperability.
   
   Solving these two issues will save millions of dollars and thousands of lives. The technology to do it is cheap and ubiquitous and paves the way for more advanced systems later on. As long as we continue to quibble about what those advanced systems need to do, we're never going to take the first step to getting there.
   
   I'll be honest, I think the issue here is involving people like Szolovits. He may be a very bright guy, but he's a researcher not an engineer. You want ideas, you go to a researcher. You want solutions, you go to an engineer.
   
   

   Reply | Report Abuse | Link to this

4. 4. torsin 01:14 PM 1/13/10

   So who will people sue? A machine? Guess there is a new American dream.

   Reply | Report Abuse | Link to this

5. 5. seamountie in reply to Spoonman 01:03 AM 1/17/10

   While getting a common language for the medical machines (and personell) is a noble idea, at the moment it is almost as inconcievable as faster than light travel.
   
   We cannot even get our every day computers to speak the same language (the OS), and users defend their choice of OS with a passionate religious furvour. All that emotional energy over a machine to entertain. Just imagine the fuss about something as important as health care.
   
   The only way to make it so would be to legislate it....and the lobbying and carrying on about that would last until the next millennium.
   

   Reply | Report Abuse | Link to this

6. 6. tachyon in reply to seamountie 12:02 PM 1/26/10

   Actually, they all do speak it, albeit using slightly different dialects: Web browsers are made to work across OSes, not to mention a tiny company that has built an OS around a browser (Hey Google!). It's not a matter of OS anyway. It's a matter of data exchange protocols. I have to agree that legislating the use of a specific protocol, existing or newfound, will be a lengthy process.

   Reply | Report Abuse | Link to this

7. 7. dancilhoney 01:41 AM 6/24/11

   Thank you for providing this info. Working out new algorithms and applying them on real-life data is an exciting and challenging task.Very helpful website.
   
   http://www.newgraduatenursingjobs.net

   Reply | Report Abuse | Link to this