It has been an interesting week! The results of the recent U.S. Presidential election came as quite a shock (and dare I say, quite a disappointment) to many of us in the United States. Just about every major poll in the days and weeks prior to last Tuesday had predicted that Hillary Clinton would win a very tight race. Several political pundits had even predicted that she would win the Electoral College but lose the popular vote. The exact opposite happened - Donald Trump won the Electoral College (fairly easily) but will likely lose the popular vote by about 200,000 votes (there are still absentee votes to be counted). What happened with the predictions? How could the pollsters have been so wrong? There are a number of theories that have been put forth - for example, some experts have suggested that some of the individuals who were polled before the election were not completely honest with who they were going to support in the election. Regardless, I think that many of us will look at polls and predictions in the future with a healthy dose of skepticism.
I have talked about a number of studies that show that statistical analyses and computer algorithms are more accurate than humans in decision making in a variety of circumstances (see "Humans versus computers in decision making" from one of my earlier blog posts). I would like to suggest, however, that there is a time and place for so-called "gut feelings" - even in health care. Narayan Kandasamy and colleagues recently published a very interesting study on "gut instinct" and "gut feelings" ("Interoceptive ability predicts survival on a London trading floor"). The investigators in this study analyzed whether interoceptive ability (the ability to detect and respond to somatic signals, such as quickening heart rate, breathlessness, "butterflies in the stomach", or sweating - basically everything that a lie detector test measures) could predict whether a high-frequency trader is successful making profitable investments. High frequency trading requires quick thinking - these traders have to make a decision in a matter of seconds on whether to buy or sell a particular stock investment. Their individual livelihoods depend upon their ability to predict the right stock. Kandasamy's team of investigators theorized that successful traders used "gut instinct" to make their decisions. Those traders who could "detect" (unconsciously, of course) an increase in their heart rate would be more attuned to their gut-level instincts - these investigators, would in turn, be more likely to make the correct (i.e. profitable) decisions on their investments. The results of this study suggested the following:
1. High-frequency traders had a better interoceptive ability (i.e. they could tell the investigators their heart rate without measuring their own pulse) compared to age- and gender-matched control subjects (graduate students).
2. Interoceptive ability predicted which high-frequency traders were the most successful (as determined by their individual profit-loss statements).
3. Experienced (in terms of the number of years on the job) high-frequency traders had a better interoceptive ability than less experienced traders.
4. The traders were asked how confident they were in determining their individual heart rates - unfortunately, there was no relationship between their degree of confidence in their heart rate predictions and the accuracy of these predictions, or in the accuracy of their decision making.
This is a great study that raises a number of interesting points to ponder further. It is clear that "gut instinct" can play a role in making accurate decisions. The dilemma here is that we really can't know whose "gut feeling" is accurate. The study suggested that the traders with a lower body mass index, lower resting heart rate, and lower beat-to-beat variability had a greater degree of interoceptive ability than the rest of the traders. I am not sure what this means, but I would surmise that those traders with a lower, more consistent heart rate would be more likely to "pick up" even a subtle increase in their heart rate when trying to make a decision (the lower resting heart rate and beat-to-beat variability, in effect, improves the signal-to-noise ratio). This would suggest that we could potentially "train" an individual to respond to "gut feelings." This is, in effect, what we are doing when we perform drills and simulations (e.g. mock codes and mass casualty exercises in health care, flight simulations in aviation, and war games in the military).
We have learned to pay attention to "gut feelings" in health care. For example, Beth Crandall published a study several years ago that suggested that nurses in the neonatal ICU respond to "gut feelings" to diagnose infants with life-threatening sepsis, often before more obvious signs and symptoms manifest. We have operationalized "gut feelings" in our hospital by teaching bedside providers to verbalize their concerns about particular patients who may get sicker and require transfer to the ICU using the term "watcher" (we published this work in the journal, Pediatrics).
So, which is better - gut instinct or computerized algorithms? The answer is probably both. I think we need to work with both cognitive psychologists and human factors engineers to develop techniques and methods to fully utilize all of the information available to the bedside provider to provide the best care to our patients - both the "Big Data" available in the electronic medical record, published literature, and Internet, as well as the information provided by so-called hunches and gut feelings. We need to identify the best way to reduce the signal-to-noise ratio of our "gut feelings" through repetitive simulation training (the more realistic the better). We also need to train providers to pay attention to their "gut feelings." And perhaps most importantly, we need to identify useful indicators of whose "gut feeling" is accurate and whose is not. Humans will never be able to pull together all of the information available as rapidly (and as accurately) as a computer can do, but computers will never be able to completely replace the human providers at the bedside. For this reason, there will always be a place for "gut feelings" and "gut instincts".
No comments:
Post a Comment