Friday, January 27, 2017

SPURIOUS CORRELATIONS -Oh My!


“Statistics teach absolutely nothing about the mode of action of medicine nor the mechanics of cure.” – Claude Bernard

Spurious Correlations, a term coined by Karl Pearson to describe the relationship between ratios and absolute measurements, is defined in Wikipedia: In statistics, a spurious relationship or spurious correlation is a mathematical relationship in which two or more events or variables are not causally related to each other (i.e. they are independent), yet it may be wrongly inferred that they are, due to either coincidence or the presence of a certain third, unseen factor (referred to as a "common response variable", "confounding factor", or "lurking variable").
Harvard Business Review cautions…https://hbr.org/2015/06/beware-spurious-correlations



Little insight is gained from the accruing enormous research. To resolve ambiguity by using mathematical probability without a tincture of skepticism sterilizes the intellect and obfuscates the truth. While mathematical probability is a human invention to ease our understanding of the world of science. “Willful ignorance entails simplifying our understanding in order to quantify our uncertainty as mathematical probability.” -Herbert Weisberg

Tylervigen.com


The scientific journals fill pages of mathematically derived studies to prove an intent, but do little to advance the cause of the physicians faced with the dilemma of treating disease.
Today quantification is the game in town. If you cannot quantify then it does not matter. Essentially the modern halls of medicine have relegated qualification of a disease process to the charred bin of history. A clinician once and still in some cases treats disease with the qualification of his or her earned and educated wisdom from experiential hindsight. The deluge of quantified guidelines inundating the landscape of medical IFTTT (If This Then That), which determines payment by the third-party insurers is circumventing the very essence of medical care of patients.

The illusion of certainty is compounded by the statistical geniuses who have little to do with medical facts or care and more to do with number manipulation to find the statistical significance. After having found the golden p-value of less than 0.05 deem the experiment, correlation, or a randomized control trial a landmark success. The quest to succeed supersedes intuition and judgment of the researcher. Today the desire for publication overrides all other questioning beasts of the mind. Studies are done for the sake of publishing and not for the sake of science of discovery itself.

No wonder the biotech giant Amgen reviewed 53 “Landmark studies” and found only 6 verifiable!

Validation data of drug target studies could only prove 14 of 67 projects.

If you look at the financial picture, the US government spends nearly $31 billion every year in science funding through NIH , which is given as research grants to academic scientists. Given the reproducibility rate of 11% (6 of 53) suggests that 89% or $28.74 Billion is being wasted. The obvious implications of such frivolous spending in healthcare costs are staggering when scaled to the entire medical industry.

Today scientific investigation considers human intuition and judgment as flawed and outmoded. Poisson once determined medical care through the lens of mathematical probability, is alive, well and wildly flourishing in the halls of scientific search. And few scientists straighten their spine to ask the question, “Can probability and statistics arbitrate the truth?”


Probability a subjective and ambiguous prospect, once an adjunct to reality has redefined itself as the objectified norm. The frequency of observable events as a hypothesis generating concept has now by the magic of quantification become “real” evidence as in medicine. The term “Risk” as one  might realize, in medicine is associated with causation, yet no precise term of the relationship between correlation and risk has a unified support, meanwhile, statistical fiat controls the issue of risk and harm. The anticipated “Harm” is exposed in this article from the Harvard Professors: https://www.washingtonpost.com/posteverything/wp/2017/01/23/repealing-the-affordable-care-act-will-kill-more-than-43000-people-annually/ The logic here relies on estimates based solely on assumptions. There are no hard facts except disparate data to prove their ideological point. So is that harm?

Calling into question the human subjectivity as a failing, quantified methods reign supreme today. Judgement, intuition, experiential reference subjugated to the quantified, computerized IFTTT norms. And the developed algorithms of best treatment are based on probability of response and the cost of the treatment. The hard truth willingly being ignored is the spark of intuition gained from the potential response of a single patient and the molecular truths that might lie beneath, rather than the quantified “logical” guideline based patient’s care. Even more dumbfounding is this concept at play in "scientific journals:"
Brenda J. Klement, Douglas F. Paulsen and Lawrence E. Wineski are authors of: Clinical Correlations as a Tool in Basic Science Medical Education Journal of Clinical correlation as a Tool in basic Science Medical Education. The article published in Journal of Medical Education and Curricular Development. In this article the authors propose the following; “Clinical correlations are tools to assist students in associating basic science concepts with a medical application or disease.”  One only needs to imagine the impact on the spongy brains that absorb these concepts and use them as the foundation for all future patient management in medicine.

tylervigen.com


While doubt and ambiguity grow uncertainty, quantified statistical inference deems to reduce doubt by applying certainty where certainty does not exist. Whereas real experimentation means a chance meeting error in the face, quantified inference on the other hand, suggests a sterilized, clean and objectified certainty. We are awash with "incremental advancements" that ebb and flow but real breakthroughs in medicine are few and far between. This exalted form of science of statistical "purity" has caused a slew of retractions in its wake. The retractions come in bigger and bigger waves, some by authors, others by peer complaints, and still others by the journals themselves. Most of these “high impact” articles have been cited in other literature and by other scientists as well. The scale of damage to the real advancement of science and medicine continues at unprecedented pace. Copied below are a few recent retractions;
5.     http://www.jbc.org/content/291/53/27434


The human mind has the uncanny ability to use logic, experience, outcomes, experimental design, minority opinion and other perspectives to grow their intuition map. From there the seeds of truth grow. The mechanized, automated, statistically quantified world of today leads to the unnatural and uncomfortable way of a flawed linear thinking. Today’s “Evidence,” as proclaimed by the statistical manipulators, remains a soft flowing sea of sand, moved by the vagaries of the winds of numerical information and/or misinformation. Panaceas abound in the form of “Coffee,” “Antioxidants,” “Vitamins” to name a few that were dismissed after being regaled in laity publications, the NEWS print and digital media. Coffee was correlated to cause cancer and then it was not.


Here are a few booms and busts related to faulty science.

Coffee causes cancer…
1.     Stocks P. (1970) Cancer mortality in relation to national consumption of cigarettes, solid fuel, tea, and coffee. Br J Cancer, 24:215–225
2.     IARC, (1991) ‘Monographs on the Evaluation of Carcinogenic Risks to Humans: Coffee, Tea, Mate, Methylxanthines and Methylglyoxal. Volume 51

Coffee does not cause cancer…
"After thoroughly reviewing more than 1,000 studies in humans and animals, the Working Group found that there was inadequate evidence for the carcinogenicity of coffee drinking overall."

Antioxidants and Cancer…
Initially Antioxidants were supposed to prevent cancer, later it was suggested that it might promote it!
Cochrane Review of the benefits of Antioxidants finds there is 1.03 higher risk of cancer with it’s use: http://www.cochrane.org/CD007176/LIVER_antioxidant-supplements-for-prevention-of-mortality-in-healthy-participants-and-patients-with-various-diseases
And studies in mice show; Antioxidants accelerate tumors in mice
Meanwhile Vitamin D still remains the science writer’s current favorite who present both sides of the benefit and risk arguments with zeal of their intent.
Current Conclusions: We found no evidence to support antioxidant supplements for primary or secondary prevention. Beta-carotene and vitamin E seem to increase mortality, and so may higher doses of vitamin A. Antioxidant supplements need to be considered as medicinal products and should undergo sufficient evaluation before marketing.


It is wise to remember Galton’s “Correlation Coefficient;” a numerical measure of the degree of relationship between two quantities, once heralded and later quantified by Karl Pearson as a measure of “partial causality,” since 1890 and through time, has been transformed by the manipulation of “quantified metrics” into “absolute causality.” This arbitrary measure of conjecture has now become fact du jour. The evidence in medical science is supposed to be based on “causation” and not using statistical generalizations via mathematical probability. At best this new "science" gives us partial causality and lends to harm for all.

No comments:

Post a Comment