Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
I'm sitting here watching PBS's Frontline's documentary League of Denial: The NFL's Concussion Crisis. which high lights one physician in particular, Dr. Bennet Omalu a forensic neuropathologist who performed the autopsy on Mike Webster. Omalu, on his own initiative decided to keep the former Pittsburg Steeler's brain for further examination although it had nothing to do with his cause of death. It was just human curiosity. What Omalu found was the first link between foot ball brain injuries and chronic traumatic encephalopathy.
Now I am not some computer Luddite, I will accept the OP's claims on their face that computers can be far more accurate at making the correct diagnosis at a higher rate than a human doctor, but my question, asked with a great deal of skepticism, is the ability of a computer to ask the right questions or more to the point, the basic curiosity to ask questions that have never been contemplated. After all, a computer is only as good as the data that it is given.
I don't know of a computer that can coax information out of a person as well as another human and that is what the best doctors are able to do, to get patients to trust them with the most intimate details of their lives, much less the ability to take subjective information, read between the lines and dig a little further to arrive at enough information to arrive at objective facts, facts that certainly can be introduce to a computer for analysis. But a computer on its own, without a physician interacting at the same human to human level that is the epitome of what a doctor patient relationship is based upon, I don't think so.
This exactly.
One day my dad went to see his internist, complaining of a little chest pain. Not bad, he thought his indigestion was acting up, and he had a long history of that. However, his internist thought he didn't look right. Following mostly intuition, he did an EKG. Dad was in the cardiac cath lab shortly thereafter and in the operating room as an emergency soon after that.
There is an aphorism that says, "When you hear hoof beats on the bridge, think of horses, not zebras."
Following a computerized algorithm might find 99.9% of the horses. But zebras do sneak onto the bridge. How good will the computer be at finding them? Will the computer order an EKG if the patient tells it his indigestion is acting up?
Also, how would a computer "practice" psychiatry, I wonder?
Methinks computers are overrated when it comes to medicine
Try reading Atul Gawande's book, Complications, written several years ago when he was still a surgical resident. His point was that in many, many cases there is no hard and fast answer as to what to do. Computer AI's require hard data to make choices. Gawande's book tells how it is virtually impossible to do that--and that frequently it is a surgeon's "best" guess as to how to proceed.
I remember one story he wrote about of a young boy brought in by flight for life for an accident. He was dying from his injuries and Gawande and a team of surgeons had to turn him on one side or the other--quickly--in order to save his life. They were not sure which way to turn him and simply guessed that the right side would be best. He wrote that had they turned him onto his left side, it would have killed him instantly, but that they didn't know it until he was turned onto his side.
Quote:
While “Complications” is full of tragic errors and near misses, the book is not intended to be an expose. Rather, Gawande asserts, it is meant to deepen our understanding of the intricacies of medicine. “In most medical writing, the doctor is either a hero or a villain,” he says, with an edge in his voice. “What I am trying to do is push beyond that and show how ordinary doctors are – and at the same time show that what they can do is extraordinary.”
In a section called “Mystery,” the author revisits cases that continue to baffle him, such as a woman whose initially unremarkable morning sickness persisted until the day she gave birth. Throughout the book, Gawande suggests that no complaint or treatment is routine. “During a surgeon’s career, there is a lot of uncertainty about exactly what to do,” Gawande explains. “You’d love for there to be clear guidelines for everything, but that’s not the case.”
The problem we mortals have with medicine, is in thinking, just as this thread indicates, that health care is simply science--when it is far from it. Cases still puzzle doctors because they cannot figure out how to cure someone, or in certain cases why they DO cure someone. Computers are all about science.
Having said that, the reason we have too many deaths due to health care is OVERTREATMENT. Doctors actually feel pressured by their patients to provide them with the latest pill advertised by a pharmaceutical company. People jump on the "new" drug that will solve their health problem or inconvenience and never listen to the sixty seconds of warning about what complications or problems have been discovered just developing the drug--let alone what might show up ten years after its introduction.
The third leading cause of death in hospitals other than the VA hospitals is medical error. VA hospitals have better outcomes and more satisfaction with service than any private health care provider. Part of that is due to the integrated, doctor developed free source software that gathers data in the VA system. It also doesn't have people moving in and out of the system as happens with the rest of health care providers when an employer changes to another carrier that uses a different set of hospital and doctors. No data is carried over. It all has to been input once again from the patient writing it down with the corresponding probability of error from the physician or hospital office---let alone the patient "forgetting" to include some items.
Try reading Best Care Ever by Phillip Longman who discovered the superior quality of VA health care while on assignment for Fortune Magazine--they didn't like his results so paid him a cancellation fee. They were wanting to paste a CEO from some hospital system on their magazine front cover. Instead, Longman wrote a comparison of private and VA health care, listing the strengths and weaknesses of both.
The science behind good health care is providing physicians with data which will guide their decisions, not in have computers make decisions in their place. We currently have robotic surgeries which have proven marginally better in some situations and actually worse in others.
Quote:
Minimally invasive surgeries such as conventional laparoscopic surgery and robotic assisted laparoscopic surgery (RALS) have significant advantages over open surgical approach including lower pain medication and decreased length of hospitalization. However, open surgery has demonstrated better success rates and shorter surgery time when compared to other modalities. Currently it is unclear which approach has better long-term clinical outcomes, better benefits.
But would you like to take a guess which one will cost you more to utilize?
Quote:
In a 2011 study from Johns Hopkins University about the marketing of the da Vinci robot, 41 percent of hospital websites included a description of robotic surgery, with 89 percent of those descriptions claiming clinical superiority. Despite this claim, only 2 percent of those hospitals made a specific comparison to open or laparoscopic surgery, which involves inserting a camera through an incision. The marketing for robotic surgery may win over more converts than the results of the surgeries.
“You start to see this is not just a trivial issue of exuberant marketing, but it is in some cases potentially inaccurate and really harmful, potentially harmful information, wrapped in the glitz and the glamor of a new technology,” said Gary Schwitzer, publisher of HealthNewsReview.org, a site devoted to reviewing media coverage of “medical treatments, tests, products and procedures.” Schwitzer has been reporting on health issues for more than 30 years.
---------------
A 2011 cohort study by the University of Modena and Reggio Emilia in Italy found that average length of hospital stays and number of surgical complications were almost identical in patients receiving traditional and robotic splenectomies (when all or part of the spleen is removed). The study of 90 patients found that the total cost of robotic spleen surgeries was nearly three times as much as conventional surgeries.
Health care as provided in the U.S. is 90% about how much is being earned. Patient outcomes are secondary. Until that concept can be somehow changed, computers and robots will simply be another red herring to cover up the dismal failure of American health care for a significant portion of its people.
Computer can get at best 30% of the info...you have to know what they are not saying and your gut...algorithms are just where you start.
Of you think medicine is just a bunch of protocols, you are way off base.
Well, as a long-term medical-field person (I designed/built the equipment), it is purely protocol, stated by an educated professional. No, I cannot state all the nerve systems in the human body, and nor do I need to be able to do so. I can read as well as the (most) of them, and make associations. I am in a different category than most (of the real) docs.... I have 24/7 hours to read, including all the technical journals, but I have no MD degree, though I don't feel it is necessary (arrogance here), as I taught my primary physicians a thing or two... It is understandable, as they are out in the world making a living, and I can just play all day...
So, to refute your comment of computers.... machines will take 90% of the marketplace... not here to deflate the MD's ego (though it IS way overblown, and you AREN'T God), but you need to consider that, granted the extreme amount of education), you are all just car mechanics working on people, on something with a complexity that is WAY,WAY higher. No offense intended.
But I don't agree with your extreme solution. I think there should be integration of doctors with computer resource and backup- a trend that is in progress today anyway. Never before in human history has so much been at their fingertips in terms of information.
I always question my doctors and do plenty of independent research on their recommendations. I research every single drug extensively before taking it. I have refused to take certain medicines due to risk/benefit. All patients are their own best advocates.
I'm quite sure people who do that are exceptions. The differences in treatment frequency ("the researchers found that patients who got the information had 26% fewer hip replacements and 38% fewer knee replacements") wouldn't be so large otherwise. Unless we have enough background knowledge about medicine, it is very difficult for a layman to be able to reasonably question a doctor. If you were very interested in cars growing up, learned about them in school for years, practiced fixing them under the eye of a teacher, and worked as a mechanic for years specializing in one model and make of the cars, wouldn't a natural, and likely correct, response be to dismiss the opinion of your customer that a head gasket needs to be replaced when you are pretty sure that it's the spark plugs that need to be changed? And would the repair process go better if you listened to that customer because that's what he wants and it's his money?
Last edited by first one; 12-16-2013 at 05:51 PM..
I work at a major local hospital for the past 14 years. When patients come into the ER, it requires communication to find out what is wrong and how it happened. Symptoms can overlap many conditions.
All possible overlapping conditions and symptoms would of course be known by the computer. It would know the exact frequencies in which each may occur, how these frequencies are modified by other factors, be up-to-date on every single study and other expert knowledge shared by humans about each of them. It could know that if this patient smokes but exercises and had a strep throat as a child and has this combination of genes, the chance of this condition increases by around 25% and know to ask about what the patient ate last night (because of a case reported at a conference that linked eating sushi to some symptom that should be considered), not just vaguely remember that smoking increases the chances and exercise lowers them, like a human might. A single interface would have the expert-level of knowledge in all specialties.
This kind of a problem does not require true AI, and it is already achievable. Try this old game made by someone for fun (and it's claimed that a better version is available) or watch IBM's Watson on Jeopardy. Unfortunately, most people haven't been exposed to any "smart" systems yet. At most, they played chess vs. the computer a couple times or were impressed that Google or Siri found some information. They are more likely to play video games, where the computer AI is intentionally dumbed down, in order to make the game more fun.
The communication itself is not a difficult problem either. Voice recognition and speech synthesis work well enough. In addition to voice, a computer can present information in visual ways.
Quote:
Originally Posted by victimofGM
People regularly lie to ER staff. The staff listen to the patient's words and body language to get to the truth.
People can learn to trust the computer instead of other humans. Near ubiquitous use of smartphones to research everything from the directions to the quality of restaurants shows that the world is moving in this direction quickly.
Last edited by first one; 12-16-2013 at 05:40 PM..
“Uniforms could be a source of contamination, but there is more concern about other surfaces around the patients,” said Russell N. Olmsted, president of the Association for Professionals in Infection Control and Epidemiology.
Sure, the computerized solutions wouldn't completely eliminate the contamination problems, but they would reduce them. Using robots to clean other surfaces would reduce them further. The article mentions nurses wearing their uniforms to grocery stores - I think we all saw that happen. It also mentions that "between one-third and one-half of health workers fail to follow good hand hygiene."
Last edited by first one; 12-16-2013 at 06:10 PM..
Omalu, on his own initiative decided to keep the former Pittsburg Steeler's brain for further examination although it had nothing to do with his cause of death.
The idea is not to replace the researchers who can come up with innovative ideas. They would still be needed. The current doctors who can do that, but are kept busy by well-understood cases and paperwork, would have more time for it.
Quote:
Originally Posted by ovcatto
I don't know of a computer that can coax information out of a person as well as another human and that is what the best doctors are able to do, to get patients to trust them with the most intimate details of their lives, much less the ability to take subjective information, read between the lines and dig a little further to arrive at enough information to arrive at objective facts, facts that certainly can be introduce to a computer for analysis. But a computer on its own, without a physician interacting at the same human to human level that is the epitome of what a doctor patient relationship is based upon, I don't think so.
I think people are not less likely to share the details of their lives with an AI than with another person. They will know that the computer has a much broader knowledge base and that it can integrate everything they share without any time limit. The computer won't judge them and won't talk about their case in an elevator. The patients won't feel ashamed about their sexual behavior or drug use in front of the machine. The AI can be quite good at asking probing questions. Yes, the best doctors might do better, but most of us don't have the best doctors and the nature of modern medicine means many patients and little time spent on each one.
Computer can get at best 30% of the info...you have to know what they are not saying and your gut...algorithms are just where you start.
Of you think medicine is just a bunch of protocols, you are way off base.
That's a straw man argument. The expert systems and the AI can do much more than follow simple protocols.
It appears that the computers can already outperform humans in cases where some effort was applied by the researchers. I might be off base but if it's just your gut opinion, it doesn't hold water. Another example: in this study, the AI easily outperformed humans doctors, and it also lowered the costs of treatments.
The machines are coming. The question is not if, but when. The anesthesiologists might not be needed during the colonoscopies soon. In a trial, this system reduced the risks of over-sedation, and it could save $1 billion per year.
Last edited by first one; 12-17-2013 at 10:26 PM..
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
Methinks computers are overrated when it comes to medicine
