Tag Archives: statistics

Medical Trends: The Holistic View

Medicine, Science, Science communication, , , , , , , , , , ,

There is clearly a lot to be learned about medicine from history.

Indeed many effective treatments can (and have) been identified not by close examination of the human body, but but the close observation of patterns in statistics.

Thus is is possible with good data, a good eye, and quite a bit of spare time, to see many of the contributing factors to disease or accidents. The famous cases include the realisation that the plague was carried by rats and that cholera was in the water. Thus was born the science of epidemiology.

I think if I was starting university again right now, there is a good chance I would have steered towards that as a profession – for it has saved countless lives, and can be done from the safety of a nice desk, replete with good coffee and a supply of biscuits. I have never been drawn to a life of tending to the ill or injured, so this would have been a nicer way to get my benevolence ‘hit’.

Alas, I studied engineering, and though perhaps I could use epidemiological methods to predict metal fatigue or bridge collapses, I am not sure that would be very useful. We engineers seem to spend much more time looking at the costs of making something, and then the price you can sell it for.

Anyway, time for the science bit…

Epidemiologists looks for patterns relating illnesses to other things: other illnesses, location, professions, exposure to animals, and many more.

There are some pretty major trends in health happening right now. For a good example, look at Hans Rosling’s presentation at TED recently. He shows, among other things, that people are living longer than ever before. Despite all the talk of the world going to pot, it seems there is an untold story – the story of how life expectancy in the developing world has been climbing beautifully for several decades. The stats tell a story of a golden age in humanity.

To go off on a slight tangent, I have to say what a pity it is that the media focus so much on the wars and tragedies. Of course, they sell sensation, so they will continue, but we humans are not used to getting news from the whole planet – we have barely got out of the trees and only left our small villages for cities an eye-blink ago. Evolutionarily speaking, our fears were programmed in a much smaller environment where news did not travel very far – the story of a death would be significant because you didn’t know very many people. Nowadays we get more news and we also know far more people because of the world of celebrity (blame the media again), and because we are so ‘networked’ (its the ‘new’ media this time) we also know a huge number of people by association. Thus we receive bad news far more often and tend to overvalue its direct threat to us.

Now let me get back on track. We are living in a golden age – better nutrition, cleaner water, the understanding of the theory of germs, and of course, advances in medicine (think vaccines, think penicillin, think surgical methods). We have benefited hugely from a better understanding of how the body works and of how fungi, bacteria and viruses work.

The activity is higher than ever on countless fronts: dementia,  HIV/AIDS, epilepsy, stroke, heart disease, and so on… but what about the big kahuna? I refer of course to cancer.

Well it has not been cured. The ‘cure for cancer’ has long been held up as the iconic challenge to medical science. Only trouble is, the challenge is flawed. There is no one ‘cancer’ – there are many different cancers  – and the little bastards are all subtle and complex – and even if you can kill one, killing it will often kill off something else, because cancers are not as alien as say viruses – they are in fact our own cells turning on us.

So rather than looking at cellular function and cunning ideas like rna interference, what can we do with epidemiology?

Cancer death rates are not independent of the death rates from other causes...

Yes, cancer is not an ‘epidemic’ – we are not studying its spread, but we can certainly study correlations and seek causation (think smoking tobacco or working with asbestos). Smarter people than I are already poring over this sort of data, and there is much hand-wringing nowadays because the ‘easy’ causes have most been found and now we are looking at the weaker correlations, where the link is not certain, or where the sacrifice to benefit ratio is unclear. Think barbecue meat, think E numbers and so on.

But I don’t want to talk about that sort of cause. What I am wondering is relates more to age…

Cancer is somewhat a statistical process – it may arise as a random mutation, which, as fate would have it, is also bad for one’s health. Many mutations result in no effect or perhaps cell death or perhaps just reduced function of the cell.  There are very few mutations that actually allow for continued (and sometimes rampant) growth and macro level harm. As the mutation is a random event, the chance of occurrence will depend on the number of ‘cellular events’ that occur in a lifetime – this is determined by two factors, the frequency of the events – and the length of the lifetime.

Now add to that randomness the fact that many cancers are slow growing – they may take too long to harm or kill you and something else gets you first.

These two factors together go to show you that the longer you live, the greater the risk of cancer development. Add to this the probable third factor that older cells are more likely to go haywire, and you can easily see why cancer is more commonly suffered by older people.

Does this mean that you risk of ‘catching’ cancer ‘today’ is less if you are younger? Well yes, if your cells (or immune system) are in better shape, mutations may be rarer and you may fend off some that do occur – however, your chance of ‘having’ cancer (rather than ‘getting’ it) are accumulative with age, so this is a very strong factor when looking at screening (looking for cancers that already exist). It is often only worthwhile screening for cancer in older people where the ‘hit rate’ makes the costs (false positives) justifiable.

This age affect is well known, but I am wondering if another factor may be throwing a curve ball into the stats – the longer lifespan of people generally.

As some types of cancers are being treated more effectively (prevented/slowed/cured), and as death from other causes (heart disease, pneumonia, tuberculosis, etc.) drops, does that not give cancer more fertile ground to wreak its havoc?

In other words, will curing other ailments, to some extent, tend to push us into the waiting arms of cancer?

And if this is already happening, perhaps the cancer death rates, hiding underneath the massive advances there may actually be an  underlying increase in death from cancer due to the increased survival of everything else.

———————

So! We all die of something eventually. I guess the question medical science now needs to ask is:  what is the best way to die? Should we be saved from one death only to have another? Will cancer rates start to creep up again as advances against cancer slow and lifespan continues upward? Time will tell!

The Statistics of Fear

Evolution, Mathematics, Science, , , , , , , ,

By all rights we should be rather scared of cars. We should regard them in much the same way we would regard a tarantula on the bedsheets or a short stroll on a tightrope. Yet we aren’t. Curious. Cars kill far more people than lava flows, piranhas or anthrax.

Perhaps they are too new for evolution to equip us?

Now if we look into the stats you realise we should be even more terrified of crap food, and the word “sedentary” should be a glare-inducing swear-word. If our bodies knew what was good for them, we would get grouchy after a day without a vigorous walk, and start shaking weakly after two. Sugar, so long sweet, should soon evolve to be utterly disgusting. Not soon enough for some.

Family Tree Nonsense

Evolution, History, Mathematics, Science, Science communication, , , , , , , , , , , , , , , ,

For many, learning about their family tree can be a real joy and pleasure.

Realising that a long distant grandfather was in the guild of barber-surgeons, or that you have a criminal or judge or, heaven forfend, someone famous in your lineage, can be a real thrill, and provide you with significant ego-boost, or perhaps a nice feeling of belonging.

The reason why it’s so often a positive experience, is due to an interesting mathematical oversight.

What do I mean?

This fractal can teach us something about ancestry: we have more ancestors than we tend to suppose...(image credit: Northwest Liberty School http://nwls.us/)

This fractal can teach us something about ancestry: we have more ancestors than we tend to suppose...(image credit: Northwest Liberty School http://nwls.us/)

I mean that people can read anything they want from a family tree, and this is made easy by the exponential nature of ancestry.

Would you find it remarkable for someone to say they were directly descended from Isaac Newton, Henry VIII or even Jesus? Would you think more of them?

What about people who say they are ‘from’ somewhere? – “my family originally come from Brittany…”, or “my family fought in the  revolutionary war…”

Analysis…

Now anyone who knows about the maths of ancestry, knows that there is actually very little remarkable about relations with famous people, especially from a long time back. If you have two parents and four grandparents, eight great-grand parents and so on, you can guess the numbers get big quite quickly.

The American revolutionary war was around 230 years ago, so perhaps eight or nine generations[1].  Nine generations back we had perhaps 2^9 (or 512) ancestors, and their folks were probably still around you can add them to the mix (another 1024)  giving over 1500 relatives, all swanning about somewhere in the world around the time of the war.

OK, you might want to reduce the number a bit due to some folks appearing  multiple times in your tree; (yes, in-breeding happens to all of us), but the number is probably still well in excess of  a thousand.

So, with over a thousand ancestors around at the time, the chance of having at least one involved in the war is pretty darn good (especially if you are were born to US citizens). I would argue that for anyone who can trace back three generations (to your eight great-grandparents) in the US, it would be far more remarkable if they didn’t have ancestors that fought in the war.

As you get further back in time, the numbers get more serious. A thousand years back , or forty generations, the straight maths gives 1 099 511 627 776 ancestors. Of course, this is impossible, as there were not enough people in the gene pool; the real number is clearly much lower and this is due to our old friend, in-breeding – where the family tree morphs into more of a family ‘web’ and involves the majority of the (breeding) population of your “gene pool”, the group that share enough in-breeding to behave somewhat like a super-organism. Where cross flow of genes between parts of the pool becomes retarded, (most usually by geographic barriers) the pool may divide and racial difference may develop.

Of course, we live in a time of great ‘connectivity’, and the US is a great example or a melting pot, with a very ‘open’ gene pool. This means that statistically, the chance that all 1000+ of one’s ancestors were around in the revolutionary war is hopelessly optimistic (unless there were special circumstances, like a closed community with a high degree of in-breeding, as may be the case with some religious groups).

So basically, anyone who says their family is all-American, “since the revolution” is being highly selective in their analysis.

Of course, western society does tend to invest much importance in the male line – which is far more specific – and would only give a couple of  chaps alive in ~1780, and if you can indeed prove this then the claim may be considered more interesting.

However, the argument that the male line is more important in some way (such as in the forming of character, or of any particular heritable trait) is pretty unconvincing. So even if you can trace a direct male line to Isaac Newton, this is no guarantee that you will pass your physics tests! Any advantages he had, will have been diluted by the 16,000 or so other folks who contributed just as many genes.

The male and  female lines can actually be traced (using mitochondrial DNA for the female line and Y-chromosomal DNA for the male line), but though this makes it easier to trace these ancestors,  it is perhaps still unwise to assume this line is more important than the thousands of other ancestors.

In the case of the USA, there is another factor, the large family sizes, and the resulting high population growth rate. The population present during the revolution have, by all accounts, been very fruitful. That means that even if you could trace your male line right back to, say, Thomas Jefferson, the chances are, you are not unique.

The Opinion Bit…

Hard-earned privilege...
Hard-earned privilege…

I am constantly annoyed by selective analysis of ancestry. I hope that the above simple illustrations alert the reader to this trickery, or at least confirm the reader’s suspicions (or convictions) that much of this is wishful thinking. What is most important to our own ‘value’ in the world is surely what we ourselves decide to do, not what our remote ancestors may have done.

However, I cannot deny that family research is still hugely interesting, even if what it really confirms is that we are all brothers and sisters, and none of us is superior due to our ancestry.

Don’t even get me started on so-called “royalty”!

[1] How long is a generation? http://www.ancestry.com/learn/library/article.aspx?article=11152

Unproven medicine : an alternative name for alternative medicine

In the media, Medicine, Science, Science communication, Skepticism, The scientific method, , , , , , , , , , , , ,

“Alternative and complimentary therapies”. They sound so nice. So warm and fuzzy. Surely they augment the cold clinical scientific approach to regular medicine, and have a more holistic approach catering to the soul and spirit as well as the flesh?

I argue not. Hear out my logic…

Any treatment that has proven to provide reliable benefit, is automatically added to the canon of ‘western’ medicine. Therefore the only treatments left available for ‘alternative’ to claim, are those that are unproven, or worse, treatments known to be actively harmful.

Promoters of alternative medicine will argue that western medicine is still woefully weak, and not tuned into holistic and spiritual matters and that such things defy proof. This is clearly claptrap. If you do a well designed double-blind, placebo-controlled test of an ‘alternative therapy’ and the outcomes are no better than for the placebo, then the participants who got the treatment are no better off, spirit or no spirit.

I personally prefer the sort of benefits that can be detected!

How did this situation come to pass, where unproven medications have such a grip?

I think there are three main ingredients:

  1. People make money from other people’s fear (in both western and alternative medicine) and that causes folks on both sides to hide or twist the facts – and also erodes the public’s trust.
  2. The fact that complimentary medicines do actually offer benefits – the well-known benefit of care and attention and also the benefit of the placebo effect – muddies the waters.
  3. It is however the human weakness of putting far too much value on anecdotal evidence that assures the future of unproven medicine.

I think that people who understand this do a disservice to our communities by giving this bad medicine the label ‘alternative’ or ‘complimentary’, so I would like to propose the term ‘unproven medicine’. I would however welcome some more lyrical suggestions!

Winning the toss

Mathematics, Science, Sports, , , , , , ,

WARNING: Please do not even try to read this unless you are cricket fan. If you are not it will only irritate you 😉 thx.

============

Test Cricket: Australia vs. South Africa: 6 tosses to ZERO, nada, nought. Time for a change in the rules.

by Jarrod Hart

============

Winning the toss in cricket can be a very significant advantage, and lead to unfairness. Let me explain…

Firstly, I am going to admit that I am an avid cricket fan. And being a nerd too, this makes me doubly susceptible to a love affair with sports statistics.

I probably learnt most of my maths skills working out how many runs Graeme Pollock required to improve his season’s average, or calculating the run rate required by Clive Rice’s Transvaal team to win the Benson & Hedges night series final at the Wanderers.

So what is my problem with the toss?

In many sports there are environmental factors that skew the game – the sun shining in your eye when you serve, the dew on the putting green, the wind behind Jonny Wilkinson, and so on.

A cricket pitch is no different. Being composed of sand and grass, it is not wholly predictable – it is also prone to evolve over time.

It can therefore be a big advantage to bat first, or perhaps to bat second. The skilled captain can often tell what the pitch will do from looking at it. To make things fair, a coin toss is used to see who gets to choose who goes first. Fine. Over the course of years, all teams will win some and lose some. However, with test matches taking five days each, the teams only play each other once every few years, so the ‘fairness’ may take a generation to arrive. 

The stats are clear. Winning the toss helps. Get the latest stats from wikipedia; at the time of writing 34.7% of toss winners had won the match. 30.8% of toss losers had won. The remainder draw/tie.

Some people will still say, oh, that’s not too big an effect, less than, say, the home advantage in football. Yes the numbers aren’t too dramatic, but what no-one seems to be pointing out is that that 3.9% difference is actually coming from a somewhat bigger difference in a smaller subset of the matches.

For example, many cricket tests are one-sided. That is to say, the favourite wins. In cricket upsets are fairly rare (rarer than in football for example). Draws yes, but complete reversals are not that common. This means that there have been lots of outclassed teams winning the toss (statistically, even the most outclassed team win the toss half the time). They have of course, mostly lost.

Secondly, there have been matches where the pitch really was constant. That is to say the toss didn’t help.

So we are left only with the matches between evenly matched teams on pitches that change. The 3.9% positive toss effect must be a stronger effect coming from this subset of the results.

I also have anecdotal evidence (all scientists, wince now). I have often watched a match, where the first team has scored 600 of a flat pitch, and then seen the second team face a ball that suddenly stays low or cuts around. We have recently seen some almighty thrashings, some by an innings plus. These by teams that were, just before the match, considered fairly equal.

Now we come to the present time. The #1 and #2 teams in the world have been locked in battle for several months. I am of course referring to the ongoing test series between South Africa and Australia. And Australia have just won their sixth toss in a row. They have won the last three matches, on ‘evolving’ pitches, and of course I am bitter, of course I am looking for excuses. But I am rational enough to see unfairness when it strikes.

I was not going to write about it. I knew it would come over as whining. Until a friend pointed out a simple solution: they could have one toss at the start of the series and then alternate the choices for the remaining tests, thereby preventing one team heaping the unfairness too high.

Today’s sixth win in a row for Australia was too much. We need a change in the rules. The South Africans have lost their #1 ranking to the toss, and it stinks.

The last 6 tests:

Test no. 1899 Toss won by Australia, match by SA (In Australia)

Test no. 1902  Toss won by Australia, match by SA (In Australia)

Test no. 1904  Toss won by Australia, match by Australia  (In Australia)

Test no. 1910 Toss won by Australia,  match by Australia (In SA)

Test no. 1913  Toss won by Australia,  match by Australia (In SA)

Test no. 1916 Yes, Toss won by Australia,  match ongoing (In SA)

Foot-note: Being a cricket stats nerd, it was of course sad news to hear that Bill Frindall, a peerless cricket analyst had died. He acted as the eye of a the nerd-storm that has been raging for years, about which most of the world was blissfully unaware. His death has left us all without bearing, little squalls in the night.  CricInfo is not not quite the same. Yes, it’s rammed full of passionate staff, many of whom are nerds and scholars of the game, but it seems to lack the Frindall touch. I ask you this: who will care about this problem with the toss in this new world order? Bill would have.

Skeptical society, the logical next step from secular society

Science communication, Skepticism, The scientific method, , , , , ,

Yes, it’s true, I’m one of those science nerds who thinks that a good scientific understanding of the world should underpin government. And education, healthcare, the law, etc.

However, I have set myself up for disappointment, because our society doesn’t work like that. It is simply much easier to use anecdotes to sway opinion, to spin data, and to manipulate with a vast arsenal of marketing tricks.

Politicians, salesmen and journalists all know that the full details (of anything), with all ifs and buts included, will not make a catchy headline or slogan, will not catch the eye or tug the heartstring.

Emotions matter more than facts. People vote with their hearts not their heads.

No amount of simple campaigning for ‘better conduct’ by will ever make a damn of difference, as the causesand incentives remain. To move on, what we need is a society that thinks for itself. A skeptical society.