Understanding Charles Darwin and his work

What exactly was  Darwin’s work about?

Charles Darwin’s work had two distinct aims: first, to demonstrate the fact of evolution (a history of life regulated by “descent with modification” in otherwords, complicated entities evolved from earlier simple ones); second, to advance the theory of natural selection as the most important mechanism of evolution.

Ok, so what’s this natural selection? how much of its random (by chance) & how much is due to design ?

       Simply put natural selection comes into play when a variation of an existing trait increases its frequency in population by giving a reproductive advantage that is heritable through the transmission of genes. In other words, (variation + differential reproduction+ heredity) of trait = natural selection.

       Ok, let’s simplify this with an example. Let’s imagine a group of beetles that are green colored living in lush green forests. On first look one feels like they were designed to be green in color to be camouflaged against predatory birds. Now lets see how natural selection works. First step is, variation and this where chance plays a part if it all it does in natural selection. So long long back let’s assume there were all red colored beetles. During different generations various mutations occurred in the genes (alleles) coding for the color of the beetle. Some of these mutations became expressive (homozygosity / dominance of allele /genetic drift) and caused various colored beetles. Let’s assume purple, brown and green colored beetles were formed. So now we have variations of an existing color.

          Second step is differential reproduction and this is where whatever element of chance that came in during generation of variation gets eliminated. In this example, red, purple and brown colored beetles tend to get eaten by birds more and survive less in numbers than green beetles do. So the green color trait enhances reproductive success. The element of chance is eliminated because if there wasn’t a green colored variety of beetle there wouldn’t have been any specific variety outnumbering the other. In other words one could say it’s kind of pre destined even before green colored beetles originated that only a green colored beetle (if ever formed) will have differential number advantage in that particular forest area. But then again u can’t say it was by intelligent design because if the green beetles hadn’t come through all the other colored beetles would have lived and died in equal proportion. To summarize, natural selection doesn’t create variations (green color), but it can only act on whatever variations are there already in a population in accordance to a selection pressure (beetle feeding birds).

        The last step is hereditary and this where end product of evolution comes out. In this case, the surviving green beetles have green baby beetles because this trait has a genetic basis. The number of green beetles will heavily outnumber all other colored beetles after few generations. In due course of time, as the food available is limited, other colored beetles will become extinct and the forest will have only green beetles

            Now the key element in natural selection is that the variation of trait should enhance reproductive success of the organism. This success is postulated to be achieved by making the organism better suited to its habitat & is called adaptation which is the end result or evidence of natural selection. So in our above example the camouflage given by green color is the evidence that natural selection has molded the beetles.

           Darwin during his time had no idea about genetics.  In last century the idea of genetics and mathematical probability of mutations causing variations has been incorporated by famous trio of JBS Haldane, Robert Fischer and Wright.

What’s the controversy about natural selection among scientists?

                  There are two schools of thought regarding how evolution really occurred. The first school (neo- Darwinian / Darwin fundamentalists!) argues that natural selection alone was the mechanism by which evolution occurred while the second school argues that natural selection along with mechanisms like genetic drift also caused evolution.

                         Neo Darwinists like Richard Dawkins, John Maynard smith etc argues that every variation of trait whose frequency has increased in a population is in some way or other an adaptation. In other words if that change of trait is closely studied it will be revealed that the variation has made the organism better suited to its habitat and increased its reproductive success.

                    Scientists like Stephen jay Gould on the other hand argue that a variation of trait can become more frequent in a population by mechanisms like genetic drift also. The two key differences in genetic drift in comparison to natural selection is that firstly it’s a random process and secondly its results are non adaptive ( doesn’t give the organism any increased reproductive success ).

How else can evolution occur other than by natural selection?

                 Genetic drift in lay man’s language can be explained as random genes drifting in by pure chance and becoming frequent phenotypes. This idea was pioneered by Sewall Wright who noticed non adaptive changes becoming frequent in small isolated populations. Again same effect is noted in human communities who are religious isolates and remain relatively insular.

whats this genetic drift?

              Genetic drift has a larger effect on small populations, but the process can occur in all populations — large or small. In pure statistical terms, genetic drift is a sampling error. For example if u have a collection of 1000 coins of which 200 are copper and 800 are gold and if u pick randomly 5 coins as a sample, then there is a high possibility that u might pick 5 silver coins alone which is a sampling error (which is not at all representative of the population as the sample size it very small). On the contrary if u pick say 100 as a random sample chances of picking 100 copper coins is very very small, though not impossible. This sampling error occurs in genetic drift (mostly in small populations) by various mechanism like  differential destruction of one variation, expression of recessive genes by inbreeding, dominant allele mutations etc.

                    In recent times , neutral theory of molecular evolution by Motoo Kimurasupports genetic drift on the basis of  observation that when one compares the genomes of existing species, the vast majority of molecular differences are selectively "neutral", i.e. the molecular changes represented by these differences do not influence the fitness of the individual organism. 

                        Since this random non adaptive changes are more likely to happen in isolated small populations(ideal for genetic drifts to become expressive) they might not leave behind “missing link fossils” (due to less number and isolated pockets). So while doing paleontological studies this will show up as abrupt changes with no intermediates or “missing link fossils”. This method of evolution is what Stephen jay Gould called as punctuated equilibrium. One has to understand that natural selection can also cause punctuated equilibrium if isolated population is subjected to new selection pressures which were not present when they were merged with main population

can this be expalined in a simple example?

               Revisiting our earlier example of beetles in lush green forest lets imagine the red colored beetles got fully replaced by purple color beetles and not green colored ones over a period of time. Now the purple color gives no adaptative advantage over red color. But such a change can still occur if the variation that occurred in first place (expression of purple color phenotype) was based on an autosmal dominant gene. Now if we imagine that these populations of beetles were an insulated group with lot of inbreeding, then over a few generations purple color can fully replace red color even if it’s a recessive gene. So here a random neutral change without any adaptation advantage(Gould named such changes spandrels) has occurred

       In the same example let’s assume  that  by the result of natural disaster (like bush fires)a small group of beetles got isolated in a part of forest where they can get better food  from the high branches of tall trees. Now due to this new selection pressure any variation that makes them better fliers (like big better wings) will be favored. In due course of time this big strong fliers will come into contact with their parent group from which they got isolated and the low flying parent beetles will get extinct. But for a paleontologist studying these beetles the evolution that took place in the small isolated forest part will be almost impossible to find due to paucity of fossils. For him or her it will look like punctuated equilibrium

medical mind 3; The great deceiver; chaos theory

The Strange story of how there is order in chaos! 

Understanding chaos in labelling a "patient" 

We all know diabetes, hypertension and increased cholesterol are bad for our health. So its only natural to assume that rigorous control of these values should be good for our health, isn't it?

The ACCORD (Action to Control Cardiovascular Risk in Diabetes) study(www.accordtrial.org) was a US federal government sponsored blinded prospective randomised control study of ten thousand adults with established type 2 diabetes who are at especially high risk of cardiovascular disease. The patients were divided into two groups, one in which diabetes was aggressively controlled ( HBA1c less than 6) and other in which HBA1c was maintained between 7 to 7.9. Based on hypertension treatment also they were grouped into two. Aggressive control group (

Results which came out in 2010 showed that both the aggressive control groups had increased mortality rates. The significance was so high in the diabetic treatment group that the study had to stopped and the aggressive control arm was switched over to normal treatment. What went wrong? The answer lies partly in chaos. Lets see how..

Met department gets it wrong again on monsoon 

     Forecasting the southwest monsoon is not easy. Till date in the 137 year history of the Indian Meteorological Department, which operationally forecasts the monsoon, has never succeeded in correctly predicting the extremes. To put this in perspective, for 137 years the Indian weather office has never ever been able to predict a drought. In last 100 years, more than 85 per cent of the time the monsoon has been normal. So the chances of a normal monsoon are always high. For last two decades, the IMD has invariably only forecast “normal” monsoons despite the huge variations India has witnessed in bad years like the massive flooding of 1994, and the droughts of 1987, 2002, 2004 and 2009. So is it that the Indian met guys are not upto the mark? Well, interestingly, for the 2012 season, the new numerical model imported from the United States actually predicted a surplus rainfall.

So why can't we predict weather? 

Prediction of extreme events is a problem since models that are used for this are faulty, in other words they tend to normalise things. Weather doesn't follow a normal distribution. What it follows can be best explained by chaos theory 

What's this chaos theory?

Well to explain it first we should be clear in our mind what it isn't about. In a scientific context, the word chaos has a different meaning than it does in its general usage. in our daily usage we use the word chaos to describe a state of confusion, lacking any order . But a chaotic system is not a random disorderly system without any underlying order.  A chaotic system  can appear like its  random and disorderly though it does have underlying order in its core

 In other words chaos theory doesn't mean that complex looking systems like weather, shapes of clouds , evolution etc are unpredictable because they are affected by too many factors and hence have a very complicated mechanism underlying them.

On the other hand , what chaos theory says is that these systems have a simple underlying order which are prone to have an unavoidable element of error/ uncertainty inbuilt in them when we analyse / measure them  . When these simple systems then go on to make complex systems by repeating themselves, the error / uncertainty will also increase exponentially making the end result unpredictable. The mathematical word for repetition is Iteration. To iterate means to repeat an operation over and over.

Chaos technical jargon 

 "Deterministic, nonlinear, dynamical systems" is the correct mathematical terminology for chaotic systems.

Dynamic means any system that changes with time.

Deterministic means these systems have an underlying order in them. Again in contrast to what we mean in common language deterministic doesn't mean something thats predictable but something that follows an underlying order.

 Non linear means  that output ( end result) isn't directly proportional to input, or that a change in one variable doesn't produce a proportional change or reaction in the related variable(s).

So in short chaotic systems are ones which evolve with time ( dynamic) which have a perfectly understandable mechanism (deterministic ) whose end result as time progresses is unpredictable ( non linear).

So, What are the key features of a chaotic system?  

1. Chaos results from a simple orderly deterministic process that’s repeated over and over. 

2. It happens only in nonlinear systems.

3. The end result looks disorganised and erratic. it can usually pass all statistical tests for randomness.

4. It happens in feedback systems—systems in which past events affect today's events, and today's events affect the future.

5. It doesn't require an external noise or disturbance to make it chaotic. For given conditions or control parameters, it's entirely self-generated. In other words, changes in other (i.e. external) variables or parameters aren't necessary. So a chaotic system is not a linear predictable system that became corrupted.

6. A chaotic system is not because of any statistical error such as sampling error or measurement error. In others words it's not a linear system that's misunderstood

7. Details of the chaotic behaviour are hypersensitive to changes in initial conditions (minor changes in the starting values of the variables).

8. Forecasts of long-term behaviour are meaningless. The reasons are sensitivity to initial conditions and the impossibility of measuring a variable to infinite accuracy.

9. Short-term predictions, however, can be relatively accurate.

What causes chaotic behaviour?

In-spite of knowing the simple event/ system which caused the complicated system / behaviour, wewon't be able to predict the characteristics of the complex system due to the inbuilt error/ uncertainty in the simple event that started it . This known as the butterfly effect or sensitive dependence on initial conditions, which many consider to be the cause of chaotic behaviour

Sensitive dependence on initial conditions (butterfly effect) means that a seemingly insignificant difference in the starting value of a variable can, over time, lead to vast differences in output. Measurement error, noise, or roundoff in the data values are the usual causes of this. For eg, computers, in particular, are susceptible to sensitivity to initial conditions because of their binary system, operating system, software details, the way they are built, and so on.

So as James Gleick put it, chaos in nothing but "Order masquerading as randomness" or to paraphrase Richard Kautz , "chaos is predictable random motion!" Now that sounds like an oxymoron, but it isn't . how can one predict something that's random?.

In short, complex systems like weather pattern , stock market behaviour, shapes of trees , maps of continents , occurrence of earth quakes, shapes of clouds and ice flakes have an underlying order which will be simple. But understanding this simple elements wont help in predicting the behaviour of the concerned system.

Why should one try to understand whether a system is chaotic or not?

1) Once we understand that a random looking phenomenon is chaotic, then that means there's a discoverable law involved, and a promise of greater understanding of underlying mechanism.eg; fractals in formation of shapes of clouds, trees, shape of continents, (keep in mind though that this doesn't mean we can predict these shapes in long term)

2) Identifying chaos can lead to greater accuracy in short-term predictions and at same time we can confidently say that long-term forecasting is largely meaningless.

3) Once we have understood the basic principles of chaos theory we have to modify what we expect when we deal with these complex non linear systems. For eg; The basis of weather forecasting is looking from a satellite and seeing a storm coming, but not predicting that the storm will form

Chaos in medicine 

Just controlling one number doesn't decrease  mortality/ morbidity risk. ACCORD study is the classical example. It showed that over enthusiastic maintenance of numerical limits like glycated Hb  below 7, BP of 120 / 80 or cholesterol value less than 200 doesn't significantly reduce risk of mortality . Just cuz we know the mechanism by which say increased glucose levels, lipid levels or hypertension can harm our arteries doesn't mean we can easily predict their impact in from of a single number

Another major issue of making cut off marks like above is that these values are in fact really close to the normal mean value of population. This means that a huge percent of normal population suddenly becomes "patients". Now once diagnosed as patient, you have to medications for the "disease". So one leads to another and wide spread usage of certain drugs are started which inturn does more harm than good as the rare adverse effects of these drugs (eg; fragility fractures in bisphosphonate therapy) becomes not so rare entities .

In short not understanding chaos can deceive clinician and patient into inverting the risk benefit ratio of treatment modalities

So now it's clear why all surgical instruments are routinely checked / replaced after specific time periods even though they are working quite well. It's because we are unable to detect or avoid all butterfly wing flappings that will later during repetitive (iterations) usage through chaos theory mechanism present as a random catastrophic event

How has chaos theory changed the way we view world around us?  

Realising that a simple, deterministic equation can create a highly irregular or random system has forced us to reconsider and revise the long-held idea of a clear separation between determinism and randomness. In other words, many of what we thought were random processes have turned out to be not so random in the sense that they have an underlying order

The separation between determinism and randomness have definite philosophical undertones.Determinism is a philosophy that says that for everything that we decide or do there are conditions such that, given them, we could have decided or done nothing else. This is often contrasted withconcept of free will which assumes we have ability to make random choices free of constraints. 

Chaos shows how something that’s very deterministic in its origin can have a random appearance in the end. This concept throws light on origins of what we feel as free will

Moral of the story

So, the lesson one learns from understanding chaos is that knowing in depth about something doesn't necessarily mean one can predict its future with reliable certainty. The inverse too is true, just because something looks like random doesn't mean its random and hence there is no point in studying it!

To conclude one can say, chaotic systems though are deterministic in root cause become probabilistic as they iterate and appear as random. So its important when we deal with systems not only to know the underlying mechanisms, but also to know in-spite of their basic ,mechanisms, whether they are deterministic or probabilistic as they evolve. Physics , chemistry are deterministic sciences while  medicine and social sciences like economics, politics,  epidemiology & geographical sciences are probabilistic in nature due to chaos.

medical mind 2; A clinician par excellence ; Bayes theorem

A clinician par excellence; Bayes theorem 

Imagine a clinician who never misses even the rarest of clinical signs. Not only that, with every rare case he sees his clinical intuition improves and he never forgets a case! 

Welcome to the world of bayes theorem, the theorem that helped Allied forces to win world war by decoding enigma and thus kick started artificial intelligence, the theorem that Google uses to give you traffic analysis in maps and drive its automatic cars, the theorem that has saved innocents from being found guilty in courts and also the theorem that works inside our mind every time we make a decision. To put it simply, the seat of practical intelligence.

A simple question to get started.... 

Let’s take example of a diagnostic test that has a sensitivity of 100% and specificity of 99% (roughly true of ELISA for AIDS).   Now your patient has been tested as positive, and he wants to know what the probability that he really has the disease is?

Reading this the first instinct is to say, its close to 90 to 100% possible that he has the disease. What’s the real answer? For that you have to know the incidence of the tested disease. Let’s, say its 1/1000. Now whats the answer?

Enter bayes. The answer is close to 9%. Meaning in 9/10 times a positive test is a false alarm! In other-words its a rubbish test

Enter the bayesian method 

In Bayesian statistics we go from data towards the hypothesis. In a classical Bayesian situation, we have a new event or data in our hand. We already know about a few ways (hypotheses / mechanisms) by which this data or event could have been possible. What essentially we want to know is what’s the probability that one of these hypotheses could have caused this event / data. If we have the probability of each of these hypothesis causing this data / event, then  In the end we can find out which of these causative hypothesises / mechanisms was more probable in causing the given event/ data

In the above example, the event that has occurred is a patient being tested positive. Now there are two mechanisms / hypothesis by which the test can be positive. One is obviously if he has the disease, and second is if the test has given a false positive result. To make it simpler imagine a sample of 1000 people on which the test is done.  We know from incidence, that 1 out of 1000 will have the disease (true positive ,which is our causative hypothesis-1). We also know that 10/ 1000 will tested as false positive by the test (causative hypothesis-2). So now we have total 11/ 1000 with positive test. But only 1/11 tested positive really has the disease. Meaning about 9% probability!

Little bit about bayesian terminology 

Bayesian equation has three parts and it involves two factors, an event and a hypothesis / belief. Essentially a Bayesian process is about how the event modifies a pre existing belief or hypothesis.

So, the three parts of famous bayes theorem is,

1.  Prior, which is the probability by which a particular hypothesis/ mechanism is known to cause the given event.

2. Likely hood, the probability that that particular hypothesis/ mechanism has caused this event in comparison with other possible competing causative hypotheses/ mechanisms.

3. Posterior, the probability that the  particular hypothesis did indeed cause the event

To summarise, in a Bayesian analysis a prior (an existing belief) can get weakened or strengthened to a posterior (new belief) when exposed to an event due to the existence of likelihood (probability of conflicting beliefs)

All clinicians are Bayesians ; well all humans are so too 

Almost all of our decisions are made by using a Bayesian analysis.  It’s surprisingly true because we as humans are subconsciously doing Bayesian analysis all the time. Its not surprising hence that bayes is the engine that drives a doctor’s clinical acumen for diagnosing.

In the above example, substitute clinical test with any clinical sign or symptom. For eg; a swollen leg on one side in parts of Kerala will be diagnostic of filarial infestation while in western world venous thrombosis will be the first diagnosis. A stiff hip joint in child can be diagnostic of a rare condition called Perthes’ disease in certain parts of the world where as somewhere else its diagnostic of tuberculosis and somewhere else its just a synovitis thats self limiting.

Bayes in action without we even knowing it & a key point 

Now imagine you walking though a street and to your horror suddenly see a human body falling down from top of a multi-storey building.  here you have been exposed to an event.

What are the possible hypothesis / mechanisms that can cause this event? It can be a murder, a suicide or even something like a death penalty being executed.

Now lets see what happens to your prior belief in each of these hypothesis depending on the likelihood .If the street u were walking was one of Mumbai underworlds’ notorious dens they u assume it to be a murder, an underworld strike. Or in other words, your prior belief of murder hypothesis gets strengthened. If u were walking through wall-street in Sep 2008 u might assume it’s a share holder committing suicide. If the street was one in Taliban ruled Afghanistan u may assume it is a death penalty given for listening to music.

The event observed here was falling of a human body.  But depending on the likely hood, we make an assumption of murder, suicide or death penalty.  Or to put it differently the incidence of murder, suicide and death penalty was different in each street and hence we made the Bayesian analysis.

The key point is that with every new information added the belief gets modified. For eg, if the body looks decomposed when it fell , the suicide prior gets weakened or if its a woman, the underworld hit belief gets weakened. Thus in our thoughts we are always using bayesian method to validate our beliefs.

why most people never heard of bayes; Frequentist statistics 

Bayesian method is not the main stream method used in statistics though. What we usually hear, do and debate about in statistics come from what’s called frequentist statistics. Its interesting because, bayesian method is what our mind seems to be following in its default mode. Now lets see in short what are the basic principles of this frequentist  method

One hypothesis , many events  

In usual statistical analyses, what we are trying to look for is the probability of getting a given data if a particular hypothesis is true or false. In other words, we have a hypothesis and then we look for whether the event or data matches it.  For example when we do a study to check the potency of a drug, we take readings before (R1) and after (R2) the drug intervention. Now we analyse these data to see what the probability is of getting the data R2, if the hypothesis (that the drug has not caused a difference in the outcome, null hypothesis) is true. In a scenario where null hypothesis is true If the probability of two data R1 and R2 occurring is less than 1/20, we reject the hypothesis and say  that the drug is effective because it made a statistically significant change (p-value less than 0.05)

But we are not wired that way.... 

Straight away what becomes obvious is why we are more comfortable with bayesian method in our default thinking mode. In bayesian method we need only very few or even at times a single event or observation to get started. As far as hypothesis and beliefs go , human mind is proficient in having many of them at finger tip.

The world of p value 

In frequentist method on the other hand , meticulous and voluminous data collection / event observation is needed to get started. Its about  the probability of getting a particular  data, given the hypothesis. That is, this approach treats data as random (if you repeated the study, the data might come out differently), and hypotheses as fixed (the hypothesis is either true or false, and so has a probability of either 1 or 0, you just don’t know for sure which it is). This approach is called frequentist because it’s concerned with the frequency with which one expects to observe the data, given some hypothesis about the world. The P values in the “Results” sections  are values of Probability of a given data / event occurring if a particular hypothesis ( usually null hypothesis) is true.

Why bayes is The seat of artificial intelligence or any intelligence 

Essentially the difference between bayesian and frequentist method can be zeroed down to what decides the probability in a given observation. In bayesian method probability depends and varies according to the state of available knowledge of the person or system that analyses the observation. In frequentist, the probability depends on the frequency of repeated observations made.

We can straight away see the problems of bayesian methodology in working of human mind with the kind of biases ,  belief variations  and limitations of knowledge acquiring and retaining abilities in each of us. The same reasons make bayesian methodology ideal for AI. Devoid of biases and clouded beliefs and capable of enormous knowledge storage, AI is tailor made for bayesian and vice versa. No wonder computer algorithms are mostly bayesian

Double agents; clinicians 

As a clinician one has to be both a frequentist and bayesian. You got be a frequentist when looking for well evidenced studies, efficacy of new drugs, making checklists and protocols. Bayesian method is handy when u get down to individual decision making about patients like which tests to do, what each test infer, which intervention suits each patient etc.

No wonder, experience matters when practicing while books matter when learning! 

It can also be said as a cause to effect journey to ascertain for a given cause is the effect possible.

Bayes in court

  In 1999, Sally Clark, a lawyer who lost her first son at 11 weeks and her second at 8 weeks, stood trial for murdering her kids and was convicted. The prime argument against her was that statistically its improbable for both these deaths to be due to SIDS and hence has to be a murder. A prominent paediatrician, Sir Roy Meadow, had testified for the prosecution about Sudden Infant Death Syndrome, known as SIDS said the incidence of one SIDS death was one in 8,500 . With such a rare incidence, SIDS happening twice is even more improbable. Verdict; guilty.

We will come back to the case little later. Lets consider an example that will make things easier and show what was the statistical error in it.

Imagine a huge basket with 1000 billiard balls in it. 996 of it are white. Four are coloured of which two are red and two are green. Now whats the probability that if u have picked two coloured balls consecutively that they. Both are red?

Answer is 1/3 rd  or 33 %chance. There are only three possibilities , both being red, both being green and one being red and one being green. Now imagine you only take the data that there is a 1/500 possibility of the ball being red and try to find out the answer. You end up multiplying 1/500 x 1/500= 1/250000 and present it as the answer!

In sally Clarke case, all you need is to substitute like this. 1000 balls equals total live births. Coloured balls are the unnatural child deaths. Red balls are the murders. Green balls are the SIDS deaths.

In the case, only the incidence of green balls was taken (1/8500). The incidence  of children murdered by mothers was recorded as about (1/22,000). About 25 times less likely. So effectively the number of green balls is 25 times more than red balls. From now on u don't need mathematics to see that the probability of both these deaths being murder is significantly less than they being due to SIDS.

The sad part is though this statistical blunder was later proved and clark was acquitted she later died of alcoholism due to the stress she had to undergo

Medical mind 1 ; The Master Healer; Regression to mean

The Master Healer; Regression to mean 

Try this question. Who in the history of mankind has healed more wounds, cured more diseases and relieved more ailments than any one? Not only is there a definite answer to this question but also  a lot of daylight is there between this master healer and the next in line.

The origin of medicine, its growth and present omnipresence owe a lot to this simple fact; a good portion of human ailments; physical and mental, regresses to the mean.

Ok, whats this regression to the mean?

Its simple, most events over a period of time will have a mean like value or state. We usually know this state by the name " normal" . When we say we had a normal day, an uneventful journey or an average performance, we are evoking this mean value.

Now lets imagine a not so normal day, a bad day for instance. By saying this we are meaning the day had few things that were beyond normal in a bad way, isn't it?  Its no surprise that all the following days wont be like that or in other words in most circumstances the day following the " bad day" is going to be a "normal day " . So it has regressed to mean.

But why bother about it?

When we don't understand this regression of mean happening, we end up attributing causes to why things happen when in reality there is no such relation. One classical example is corporal punishment. When someone does something thats deserving of a punishment it means his or her action was way beyond the normal variation. What it also means is in most cases that action is not going to be repeated in immediate future. Now if punishment immediately follows the act, one might feel like corporal punishment has worked as the person's behaviour seems to have improved immediately after it

The same mechanism is in action when a skilled surgeon who had a tough case on a day attributes meticulous planning and work up to why his next cases went smoothly.

Whats its relevance in medicine? 

Its easy to see how regression to mean works in quack medicine . We call it by the name " placebo effect" . Imagine having a common cold for instance, a self limiting disease. It will be at its distressing worst u when you decide to consult the "doctor”. The self limiting disease must be near its peak effect when u take the so called medicine or medicinal advice. And in days following it, you start getting better.  So the credit that has to go to regression to the mean goes to your "doctor"

Modern medicine and regression to the mean

Whats interesting though is how much so called modern medicine is indebted to ROTM. For major part of modern medicine's history, actually till about a century ago, medicine did more harm than good. With bloodletting & putting holes in skull with absolutely no idea of asepsis and  anaesthesia, a modern medicine doctor of 19th century definitely was a person to avoid if u were sick. But, still physicians and doctors were not only consulted and revered, but also made fantastic cures!  Actually they were just snatching off the credit from ROTM.

Even today the story isn't much different!

Yeah! Its a fact. Though modern medicine has made some giant leaps in fields like asepsis, anaesthesia, immunisation etc, its still owes a lot to ROTM.

Try this fact. One would expect the life expectancy changes brought on by modern medicine to be most evident in western societies. Now, the life expectancy of a child born in 1900 in US was about 45 yrs. today its close to 75 yrs. a gain of 30 yrs. but, the life expectancy of a 50 yr old was 22 yrs in 1900, and is 29 yrs today. Just a gain of 7 yrs.

What does this say? Its reasonable to assume that in last 100 yrs we have achieved some remarkable things in mother and child health care. Interestingly this where asepsis, anaesthesia, immunisation all intersects. But what it also says is in regards to adult health our achievements have being less than spectacular to say the least.

So a 50 plus year old in 2012 with back pain, knee pain or on a cocktail on medications for hypertension, anti coagulants, anti anxiety drugs, anti oxidants, etc etc is not getting dramatically better management than his counterpart  a century back. Thank god for ROTM! Patients still believe visit to GP is doing wonders and gets cured too!!