As algorithms replaces medical professions, don't forget the social signalling we pay so much fo
--this article is inspired by the brilliant book ' elephant in the brain' -----
Medical spending is 12% of Canadian GDP. That's more than$242 billion dollars or 6600 per Canadian. In comparison, Spain and Italy both have low
ered levels of spending (~3300/person) than Canada, yet their life expectancy is the same or higher life expectancies. There is no question that medicine is critical to our health, but is increased levels of spending associated with improved survival outcomes?
To start the investigation, many researchers started look into the effects of regionally marginal medicine. Since 1969, studies (1,2,3,4,5,6) have consistently shown that higher spending regions were no healthier than other patients. In comparison, it was found that there is more statistic relevance in predicting death rates to other factors such as income, education, but not by medical spending (1). Another study studied 5 million Medicare patients across 3400 U.S. hospital regions and found that for reach extra day in the intensive care unit (ICU), patients were estimated to live roughly 40 fewer days (5). The study also shows that spending an additional $1000 on a patient resulted in somewhere between a gain of 5 days and a loss of 20 days of life.
While these studies are just correlation affects and the relationship might very well have been artificial, RAND corporation went beyond the correlation studies and investigated if medical care actually causes better outcomes. Between 1974 and 1982, the RAND Corporation performed a randomized controlled study on 5800 non-elderly adults from six U.S. cities. Each individual is given the same access to the same set of doctors and hospitals if they were in the same city, but were randomly assigned different levels of medical subsidies. Patients whose medicine was fully subsidized consumed 45% more than patients in the unsubsidized group but with NO detectable health differences.
In fact, some has shown that health care sometimes has opposite effects than we anticipated. In America, medical errors are the third leading cause of death, more than the total amount of deaths from firearms, suicide, vehicle accidents, the third largest killer after cancer and heart diseases is medical care (7). What is more shocking is that even though the final autopsy results shows that 40% of the initial diagnosis is wrong (8), autopsy rates actually decreased from a high of 50 percent in the 1950s to a current rate of about 5 perfect (9).
Despite all of the inefficiencies, medical reform does not happen, and the general public ignores the waste and errors and shows very little interest in actual medical quality. For example, patients who would soon undergo a dangerous surgery (with a few percent chance of death) were offered private information on the risk of dying correspondent to individual doctors and hospitals at a price of $50. These rates were large and varied by a factor of three. However, only 8 percent of patients were willing to spend even $50 to learn these death rates (10).
Fortunately, there has been showing excitement to push through medical reforms due to some break through in new technologies. We have consistently seem that deep neural network algorithms are now better at diagnosing specialist cases than doctors, whenever it is given the data and training required. One of the studies, published on Chex Net, has shows that algorithm can spot pneumonia better than a radiologist, and identify heart arrhythmias better than a human expert, or even identify domestic abuse from a coagulated type of data that's not available to any single medical doctor (11).
As medical care become more efficient as we replace human biases and errors with algorithm, we must not forget about the social comfort required by patients. As the book ‘Elephant in the Brain’ points out, the reason we have refused any medical reform largely lies in that a huge of medical costs is generated due to the need for social signaling. It is an evolutionary desire for a conspicuous care act made in public in order to send a social signal to people that we have physical and political supports. Therefore, the need and job market for social interactions and personal care should not be ignored while we make the medical institution more efficient.
1.Auster, Leveson, and Sarachk 1969.
2 Fisher et al. 2003.
3. Fisher et al. (2000)
4.Byrne et al. 2006.
5.Skinner and Wennberg 2000”
6.Hadley 1982
7. Martin A Makary, 2016
8. Lundberg 1998
9 Shojania et al. 2002
10. Mundinger et al. 2000
11. Pranav, etc. 2017
