The evening was a welcome cool from the day’s sun. Summer was hesitant to give way. Perhaps it had a few more thunderstorms up its sleeve or was mostly tempting Fall. I couldn’t say. It was a magical evening nonetheless. As the fading light of the setting sun caressed the tall pine trees and disappeared ever so slowly a cacophony arose from the trees. The cicadas had arisen from their deep slumber and now vibrating their tymbals were taunting their predators with their rhetoric. It became constant and sort of annoying. It reminded me of the itinerant intellectuals who pass along their confined boxed-set knowledge as the truth.
We seem to live in a world inhabited by Sophists; the Protagoras and Thales kind; forever drifting in the sea of their own making. Constantly deducting from a set of false premises. Plato would have set them free, but alas, very few ascribe to those ideals. Their drumbeat resonates in more ears with their tymbals than tongues splashed from the taste of fine wine poured from the vineyards of reality. Relativism seems to play a large part in our society. Be it moral, metaphysical or scientific. In every corner of the inhabited land a call for relativism is trumpeting the ever increasing vigilance of “All things are equal.” “There is no good nor bad, only thinking makes it so.” Hmm, where have I heard that before?
You see dear friend and reader, when we start preaching from that diamond-studded lectern and profess an intellectual high ground that is fortified only by false premise, someone, some day, with an average intellect will eventually find the key to the falsehoods.
Now imagine the falsehoods being churned out in large numbers in science. And, I am speaking strictly about Medical Science. Causality is inferred through an elaborate scheme of statistical fiat. From the simple and easily understood Kaplan Meyer Curve that showed survival in a graphic form; The survivors in the treatment arm vs. in the untreated arm of a study. We have now launched into Forest Plots, Propensity Scores, Instrumental Analyses, p-Values and other elaborate probable causality forms. We continue to draw strong inferences from weak correlations. One can reach logical conclusions through reasoning based on any premise. And that is where the fly in the ointment sits. As the term “garbage in, garbage out” is used in computer input/output results, so are the conclusions based on inferences. Statistical Inference on the other hand provides us with a probabilistic “determinist reasoning” where predictability is circumspect in most cases relying on percentages. No wonder IBM’s Watson failed to provide the correct treatment in cancer cases - a failure of AI dependent on limited and incongruous data. Similarly having a false premise and even worse, falsified variables lead to false conclusions. By simply masquerading as a statistician or being one employed to determine a firm value to take to the bank, one can herald a meaningless treatment protocol costing hundreds of thousands of shekels to prove that an occasional survivor from an n of 14 will come forth and brighten the balance sheet. The data so incurred is as flippant as the coin being flipped. The cicadas are vibrating their tymbals ever louder as average people view the data through a clear lens and say, “What gives?”
What do we do. Should we let causality be an inferred determinant from correlated inferences? Should we simply bow down to these itinerant composers of rhetoric and their daily fusillade of useless information couched as “landmark studies?” We should reason! However, knowing well, If one questions the motives of these falsehoods, moral relativism will come into play and virtue-signaling, now an art form for some, will blind you, deafen you and vilify you to shut you down. Perhaps we might consider using the correlational inference as the intermediate step and then look hard for the proof on causality? Something akin to finding the Anopheles mosquito as the vector in Malaria, or lone star tick for Lymes Disease.
This storm of falsehoods and intemperate pseudo-scientific behavior will come to an end, one day. More transparency is being demanded by those that inhale this form of information because the links between therapy and disease outcome seem to move farther apart that neither the statistician nor the scientist/author can bring the opposing factions together in a cohesive bundle. Numbers can’t pull together what common sense says does not belong. After all, adding more salt or sugar will not displace the aged and oxidized fruit to a better shelf. Medicines that are used as treatment of disease are all the more of concern. Especially when lives and billions of dollars are at stake. Hiding behind the Progression Free Survival instead of Overall Survival will simply not do. After all, by the magic of “timed evaluation” a PFS can vouch for a “markedly significant” response although overall survival of the patient remains unchanged. Meanwhile billions of dollars go down the drain and precious lives are lost for the want of time. Read the literature and I promise, if you do it with clear eyes and open mind, a tear of sadness shall flow unannounced.
So once again I get back against their (Sophists) ideal, that there are some absolutes in this world. Vanquishing disease for instance, delaying disease as the next best, or simply creating comfort are the three modalities of care in cancer medicine and most other diseases. One should not have to pay dearly in finances and lose a fortune in the hopes of living longer when the ticking of the clock suggests that time-determinant survival is not prolonged, except on the bias of variables, a code for snake oil and the fiat of statistics. Those patient’s finances are better left in the hands of the individual and the family to be used for better purposes than in the hands of wealthy bureaucrats. Free Market is free only when there are no controls in place that displace the real cost of "something." Calling the current scenario in medicine as free market is tantamount to calling the sickle and hammer flag a beacon of Democracy.
Causal inferences should be drawn from verified and validated data. If only 6 of 53 “landmark studies” could be validated by Amgen then something is wrong and twisted, in the land of numbers. And now with the Genomics at hand where thousands of genes are in play and perhaps thousands more of epigenetic forces (RNAs) are modulating that play, statistical paradise might have some more bad calls as we muddle through treatment of diseases.
Harkening back to the days when I was in my teens, I thought all knowledge about disease had been collated and complete, that there was no more knowledge to be had...
Oh man, was I wrong!
I come back to the original correlational question, “Is Causality dead?” As Dylan answered, “the answer my friend, is blowin in the wind.”
No comments:
Post a Comment