 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
Medical Forum / General / General / April 2005Are there any computerized diagnostic programs?
The world of modern medicine has become so complex that doctors have segregated themselves into specialties and subspecialties. It is impossible for one human brain to integrate and evaluate the current avalanche of medical knowledge. As a result we encounter mis-diagnoses on the part of doctors and a feeling of helplessness on the part of patients. Why is there no medical software program that integrates the entire medical knowledge base, diagnostic procedures and differential diagnoses? A modern computer would have no problem analyzing a medical problem after being fed diagnostic results and up to date data base of medical knowledge. Sure, such a program would be complex but certainly not as complex as the engineering details for a 747 airliner. Maybe there is such a thing? Can't find it in Google.  SignatureJames >Sure, such a program would be complex but certainly not as complex as the >engineering details for a 747 airliner. You underestimate it by orders of magnitude. A 747 is a complete triviality in comparison. My understanding is that this type of software has been used in limited trials, and outperforms physicians by a good margin. And it is quite straightforward to implement - as you say, much simpler than designing a 747. But there are two huge obstacles to overcome: 1. Physicians will never allow anything that cuts into their pay and prestige. Never, ever. 2. Patients would also be somewhat uneasy with such a thing. Would you feel comfortable flying in a plane piloted only by a computer, even if the computer had been shown to handily outperform a human pilot? =OCCAMMAN= > The world of modern medicine has become so complex that doctors have > segregated themselves into specialties and subspecialties. It is impossible [quoted text clipped - 11 lines] > > Maybe there is such a thing? Can't find it in Google. Yeah, but it would be nice to second-guess my doctor. SignatureJames > My understanding is that this type of software has been used in limited > trials, and outperforms physicians by a good margin. And it is quite [quoted text clipped - 27 lines] >> >> Maybe there is such a thing? Can't find it in Google. >My understanding is that this type of software has been used in limited >trials, and outperforms physicians by a good margin. Citation please. >And it is quite >straightforward to implement - as you say, much simpler than designing a >747. Flying is physics. The calculations can be programmed from books in most cases, and determined with precision in the others. Yes, a modern aircraft is a complex thing, but from a computer's point of view it is no more complex than a bicycle. You seem to be under the misunderstanding that quantity of data = complexity. It does not. You also seem to think that a simple database problem like building a machine (no matter how complex) is comparable to diagnosing and fixing that machine. I'm afraid you're wrong there. The former can be readily accomplished by computers, but the latter, hardly at all. > Citation please. OK, I went and did some work for you, at no charge. A quick poke around found a few, for example: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=1 556536&dopt=Abstract http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=pubmed&dopt=Abstra ct&list_uids=6796019 http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9 356857&dopt=Abstract http://csdl.computer.org/comp/proceedings/cbms/1999/0234/00/02340296abs.htm Notes: 1. There are not a lot of these trials run. I'll leave it to your speculation as to why. 2. It's very interesting that I did not find any studies that found that computer-based diagnostic systems did an inferior job. 3. I've been involved in the implementation of a number of clinical "decision support" systems. However, even though the performance of these is not well tested, they ought to be called "decision making" systems, as once they seem to work reasonably well, my colleagues essentially rubber-stamp the answer that they spit out. > Flying is physics. The calculations can be programmed from books in most >> cases, and determined with precision in the others. Yes, a modern aircraft [quoted text clipped - 6 lines] >> machine. I'm afraid you're wrong there. The former can be readily >> accomplished by computers, but the latter, hardly at all. Being a human is also physics, last time I checked - what's your point? 747s are a lot more complex than bicycles - 747 design requires thousands of folks working together, an unbelievably complex task. Quantity of data needed to characterize something is absolutely equal to its complexity. It can be easily seen that diagnosing disease is far simpler than designing a 747 - diagnosis is usually performed by a single person. While the physics and chemistry that make up an organism may be complex, you're badly fooling yourself if you believe that physicians give even a passing thought to this when diagnosing a headache and prescribing an aspirin - or a Vioxx, if sufficiently compensated by pharmaceutical companies. Most docs follow reasonably simple algorithms in performing a diagnosis. =OCCAMMAN= "God heals and the doctor takes the fee." - Benjamin Franklin >>My understanding is that this type of software has been used in limited >>trials, and outperforms physicians by a good margin. [quoted text clipped - 15 lines] > machine. I'm afraid you're wrong there. The former can be readily > accomplished by computers, but the latter, hardly at all. >> Citation please. >OK, I went and did some work for you, at no charge. A quick poke around [quoted text clipped - 4 lines] >http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&list_uids=9 356857&dopt=Abstract >http://csdl.computer.org/comp/proceedings/cbms/1999/0234/00/02340296abs.htm Yes, indeed, computers can do some very specific diagnostic tasks far more accurately than humans. That is hardly news. In fact, it is downright ancient news. What I was looking for was a citation to support your sweeping assertions. >Notes: >1. There are not a lot of these trials run. I'll leave it to your >speculation as to why. Please be specific and tell us why you think so. I already know why I do. >2. It's very interesting that I did not find any studies that found that >computer-based diagnostic systems did an inferior job. I'm shocked, simply shocked. >3. I've been involved in the implementation of a number of clinical >"decision support" systems. However, even though the performance of >these is not well tested, they ought to be called "decision making" >systems, as once they seem to work reasonably well, my colleagues >essentially rubber-stamp the answer that they spit out. Um.... so? >747s are a lot more complex than bicycles - 747 design requires >thousands of folks working together, an unbelievably complex task. I guess the analogy was lost on you. Yes, computers are good at crunching large quantities of data. That's why we invented them, after all. But no matter how complex you think a 747 is, the fact remains that it is a triviality compared to any mammal. Aircraft are designed and built. They come with specs and manuals. They are engineering problems, not science problems. If you don't understand the huge gulf of difference between the two, then I think this conversation is probably pointless. >Quantity of data needed to characterize something is absolutely equal to >its complexity. Really now. You can't think of trivially simple problems that involve huge quantities of data, or incredibly complex problems that involve tiny quantities of data? Not a single one? >It can be easily seen that diagnosing disease is far simpler than >designing a 747 How beautifully sophomoric. Look, cut through the crap and answer this one question: Then why isn't some large corporation fabulously wealthy from having done so? Do you think they lack the necessary talent, motivation, investors, or cash? > The world of modern medicine has become so complex that doctors have > segregated themselves into specialties and subspecialties. It is impossible [quoted text clipped - 11 lines] > > Maybe there is such a thing? Can't find it in Google. Many computer programs have been written, but none that perform well enough for routine use. Medical diagnosis uses pattern-matching skills much more than database analysis, and humans are typically much better at pattern matching than computers. The engineering details for a 747 is not only not as complex as medical diagnosis, it's not even the same realm of problem. A better example would be computer-based visual recognition programs. These are far simpler but at least in the same space (pattern recognition). As far as I know, there are no computer programs that would be able to identify your spouse 100 feet away at dusk and walking away from the camera with different color hair from that previously seen and wearing new clothes. Most humans would do quite well with this much easier problem. SignatureDavid Rind drind@caregroup.harvard.edu
|
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.