Second Opinion Should we rely on a computer?

by David Orenstein free-lance writer

Overheard at a recent (but entirely allegorical) party.

Dr Tom Technophile: Hey there, Sandra. I’ve been thinking more about our ongoing discussion— you know, whether information technology is making medicine more humane and more efficient. What got me thinking was this inspiring blurb in Science (Dec 12, 2003) about how CT and MRI scanning could replace autopsies. Investigators at the University of Berne in Switzerland are gathering vital evidence without slicing open and further disturbing the victims’ bodies.

Dr Sandra Skeptic: I saw that item. Pretty interesting. What did you make of George Lundberg’s comment to Science that these so-called “virtual autopsies” miss important sensory cues such as color, feel, and smell? Actual is better than virtual if you want a complete sense of what killed someone. That’s the point you always seem to miss, Tom. We miss a lot of opportunities when we rely too heavily on machines. I bet you still think evidence-based medicine is perfect.

Dr T: I don’t mind a system that identifies people with chronic conditions and then helps the physicians and patients focus on better care. After disease management tools have helped identify that a patient has a problem, things like monitoring, behavior modification, and communication are important. And, yes, I like the fact that backing up this process is a sophisticated analysis of a broad array of clinical databases. I don’t mind intuition in medicine, but we should embrace ways to analyze the evidence we collect.

Dr S: Maybe so, but the peril of relying on algorithms and databases is a real one. The AMA has justified concerns about DM. When a DM regimen is applied too rigidly, it can fail to address individual patient circumstances and even can substitute for the physician’s judgment. That is something no computer-based system should ever do. The term “cookbook medicine” isn’t bandied about here for nothing.

Dr T: I certainly don’t support limiting the choices a physician and patient can make. I’m saying that information technology, like the relational databases used for DM, helps us sort through reams of data that are too much to sort through “by hand.” You talked about missed opportunities. Ignoring the patterns in data is one of those.

Surely, there’s a happy medium where, instead of relying too heavily on technology, we let it help us make better decisions at, as an added benefit, a lower cost. Look at automated Pap smear screening, which now has FDA approval. Those systems, from companies such as TriPath Imaging or Cytyc, let clinicians focus on the slides that most urgently need a human professional’s analysis.

Dr S: Wow, that whole discussion goes back at least a decade. I remember that some pathologists were skeptical about those systems even when they were proposed simply as tools for rescreening slides that already were manually reviewed. Apparently not everyone was comfortable substituting artificial intelligence when real intelligence—and experience—was available.

Dr T: The point is that there is a choice. Manual screening still is an option, as is the assistance of automated screening. A lot of tests and trials have been done of those systems and they have performed well.

Dr S: It worries me when the technology becomes so revered for its efficiency and low cost, that it ends up threatening or even replacing the judgment of physicians. The temptation to save money may seduce some into believing too much in machines.

Dr T: Of course, I’d never advocate a reduction in the quality of care. We can take advantage of information technology without letting it ruin us.

David Orenstein is a technology and business writer in Silicon Valley. To learn more about a technology topic in Computing Care, e-mail him at davealli@comcast.net.