Need for statistical dumping ground

Something that I would like to see is a massive medical study based just on generating statistical results without testing any hypothesis.  Just get as much information relevant to medical history about as many people as possible about diet, types of infections, injuries, genetic background, toxin exposure and anything else, dump it all into a supercomputer and see what correlations come out. 
 
Ordinary one hypothesis studies are wasteful in some ways.  If you have the subject there for analysis, looking at one field is inefficient.  Scientists like control groups, and these are important for testing a specific hypothesis, but that is it.
 
I was recently reading about an interesting idea about Alzheimers, that there is a high correlation with type 1 herpes and those with a certain gene that is vulnerable to that type of infection. 
 
This is interesting because it is widespread, with supposed estimated exposures of 85-90% of the population.  That is interesting because it implies that most people with exposure don’t have significant breakouts.
 
Problem with this study: as usual, the statistical size was extremely small.   There were about 100 brains studied, about half with Alzheimers and half without. 
 
That may mean that a very few people could make a radical change in the statistical analysis. 
 
The herpes connection is an interesting idea, one where both the proponents and opponents are equally missing out on the possible significance. 
 
Herpes never really goes away.  That means that, on some level, the body is always fighting it.  The description of this as "dormant" is probably misleading.  More likely is that it is hard for some reason to kill or fix the very last infected cell.  That may have something to do with nerve cells being a little more permanent than, say, skin cells and the body won’t like destroying its’ own nervous system to root out a virus. 
 
What else do we know about chronic viruses?   Take a look at hepatitis C for example, which can over time lead to scarring of the liver because of the body attacking the virus
 
So what happens if you think of Alzheimers as being something like the equivalent of hepatitis but in the brain?
 
A lifetime of beating down a virus can potentially leave an organ scarred.  More pervasive illnesses are more likely to do this if damage accummulates over time.  On the other hand, they don’t have to be the unique source of causation (one of the problems with the herpes theory being that not all Alzheimers patients have it) as other diseases can potentially cause the same reaction.  
 
This could also tie in with the issue of low glucose in affected brain areas being associated with Alzheimers.   Infected areas could be isolated with less blood flow to reduce spread of the virus.
 
Of course these are all just hypotheses and could be nothing more than statistical noise. 
 
It does raise a further interesting question about symbiosis.    We tend to think of viruses only from the point of view of what damage they do and that could be obstructing understanding.   If it is true that type 1 herpes doesn’t do much to most people  (and note Hep C has little effect on most people), is there adaptation to it, or do some of these viruses serve a function of hitting genetic enemies without hurting friendlies?
Advertisements

Methodological problems with climate

 
There was a raging debate around ten years ago in climatology about whether the earth was actually warming or cooling, and the debate is not as straightforward as is depicted online.
 
The raging debate had good reason.   The selection for data points is awful for spot temperatures on the planet because they are overwhelmingly near highly populated areas.  Quite apart from global warming issues cities have lots of activity in them that can warm the atmosphere.
 
Satellites have a bigger scope and are able to view and take readings from the entire earth and so cover a lot more area than the poorly spread sensors at the surface.  On the other hand, they do not measure the temperature of the surface.  They measure the temperature above the surface. 
 
The temperatures of some of the layers of upper atmosphere are cooling.   The issue then becomes, are they cooling because the earth is cooling, or are these layers cooling because of greenhouse gases that keep more heat in?
 
Something that ought to be looked at is using the satellites to measure all wavelengths of light going in and all wavelengths of light going out.  If you can measure that you can calculate net energy retained, stripped of semantics or theoretical explanation. 
 
There may of course be additional sources of heat, such as the effects of electromagnetic fields- for example making the atmosphere or earth’s core move and generate heat- or gravitational fields- for example the tides caused by the moon probably translate into heat by causing motion- or chemical reactions etc.   A good starting place though is to look at all radiation in minus all radiation out to calculate the net energy balance in one area of enquiry. 
 
Of course solving the question isn’t nearly as fun as bickering for 15 years.

FDA does something useful

 
Well, there’s a first time for everything.  
 
Food nutrition is the area that ought to be focused on first in the health care debate.   This plan, that plan, private, public, blah, blah, blah.   Why not start by dealing with the elephant in the room that nobody is talking about, things that we can do right now regardless of which plan is adopted to help solve the problems.   Do the painfully obvious things first and then go back to your rhetorical flourishes and psychodrama. 
 
Go through a supermarket looking at lists of ingredients and see how many things haven’t been junkified. 
 
Too much sugar, too much salt, ingredients you can’t pronounce that take years off your life, known carcinogens, trans fats…
 
Tax the bejesus out of all of it and you help solve two problems, without a downside.  You improve health and bring in revenues at the same time.  I don’t like the idea of bans except in the clearest of cases because the idea isn’t to turn the nation into a police state run by health authorities where anything enjoyable is illegal.  Put taxes in and the market will help correct the problem.   That doesn’t mean that you can’t have the occasional ice cream. 
 
I don’t consider undermining the sugar, wheat and similar industries a downside.  
 
Something like taxing things with a glycemic index of over 50 at a rate of 5 cents per gram of carbohydrate and over 70 at 10 cents per gram of carbohydrate would do a lot to set things straight.
 
Tax sodium chloride at ten cents a gram.   
 
How much would that save in terms of health care costs?  It is difficult to calculate that and the bureaucratic method of analysing everything to death and pulling numbers out of the air that are highly speculative won’t do much other than cause delays in implementation.  Suffice it to say that there would be an expected benefit.   Large changes in policies always bring some unforeseen results and you may as well get on with it and see what they are. 
 
 
 
 

Data vs. Information and the plausibility trap

I ran into this interesting little article about a supposed association between post traumatic stress and survival post surgery:
 
 
What is most interesting is that the statistics are given primacy although the other information which is more imprecise is actually more informative.
 
From the pool of people analysed, 129 had post traumatic stress prior to surgery.   Of these 8.5% (meaning 11 I assume) died within a year while in the non PTSD group 6.8 percent died.  That percentage would be equal to 9 out of 129 people dieing.
 
Put bluntly, a swing of 2 people out of 129 is irrelevant.  The sample size is small. 
 
Further, the studies link surgery and death, but do not link the surgery TO the death.  The report just says that they died within a year but not of what. 
 
Statistics with small numbers must be used with great caution.   The average outcome of 10 tosses of a coin will be 5 heads, 5 tails assuming the coin is balanced.  But the chance that of a given 10 tosses exactly 5 will be heads is 252/1024, about 25%.  The odds that 7 tosses would be heads is 120/1024, about 12%.  If 7 tosses came up heads it would not follow that the coin was unbalanced.  I would add that if 5 tosses came up heads it would not follow that the coin was balanced either.
 
To make a well publicized hypothesis of a 25% increased death rate on the basis of the deaths of two additional people is not particularly responsible. 
 
On the other hand the information that is difficult to massage into meaningful statistics is frankly more useful, like that PTSD people are more likely to drink, be overweight, etc.    This is more imprecise because unless you introduce a question begging line- say that 15 drinks per week is not abuse but 16 is, that a BMI of 25 is fine but 25.1 is obese- it is hard to get clear cut, unfuzzy statistical information.   That doesn’t mean that the information is without value.
 
There is another important lesson here, about the dangers of plausible theories.   Heretical theories tend to lead to better science because they will draw attacks such that they are unlikely to stand unless true or close to true.   On the other hand plausible theories can lead to junk science as they may be skimmed over as trueisms.   The above theory is one that is inherently plausible.   You would expect most physical or mental disabilities to be associated with an increased death rate.
 
There is also the red herring element in that one would expect in the abstract that those with a disability would be expected to have higher death rates than average even without throwing the issue of surgery into the mix (there is also a suppressed fuzzy element here about what constitutes major surgery and if hospitalization is all that is required then imagine the difference in longevity between say a broken leg vs. open heart surgery).    A baseline of the increased risk of death with PTSD independent of specific risk factors ought to be established before moving into more specifics.
 
 

Berlusconi- a disgrace for western democracy

Given that Mr. Berlusconi was a member of the P2 criminal organization I am amazed that he was ever in public office, let alone the Italian Prime Minister.  There have been various corruption trials, attempts by Mr. Berlusconi to suppress the press, attempts by Mr. Berlusconi to make unconstitional laws to protect himself and other Italian politicians from criminal investigation and prosecution, and a general track record of misconduct going back nearly 30 years.
 
Removal from public office? When are the Italians going to put this guy in jail?
 
I’m surprised that foreign leaders will receive him.  I wouldn’t give his office recognition.   Dealing with third world leaders with a spotty background is often necessary, but we shouldn’t be dignifying such a selection from one of our own.

General effect of innoculation on the immune system needs to be studied

In the course of the present debate on whether to innoculate against the swine flu or the regular flu, an interesting issue arose that needs to be looked at for all innoculations.
 
Apparently innoculation against one type of flu increases the chance of getting the other type of flu. 
 
That has implications for how we should measure the success of innoculations.    If innoculations can reduce resistance to other types of infections we need to study how often such an effect occurs and not limit that to studies of flu. 
 
The way in which immunizations could reduce immunity against other diseases goes like this: the immune system is highly adaptive and has finite resources.  There is some automated way of determining threat probability and allocating resources based on encountered threats.   The implication is that a massive, immunization style flood of an infectious agent may cause the immune system not just to create more of a response to that infectious agent, but also by diverting resources away from either other similar infectious agents or from other immune programs generally.   In the event that any immunizations are additive, it should be looked at whether excess immune response may be taking resources away from other areas of health which may result in other health problems, i.e. does the immune system compete with any other bodily systems for resources. 
 
I’m not suggesting that immunizations are a bad thing, although they may be.    Our immune systems have not had sufficient time to evolve to meet modern challenges and they may need a boost.    Modern societies bring plagues because diseases go around and around and get to mutate through exposures to billions of people.  In more primitive societies with  more limited interactions and more limited exposures to populations, reducing the chance for mutation, exposure to something like a new flu may have been more like a once in a lifetime rather than a once a year process and the immune system would probably be adapted accordingly, with greater resistance to things like bacteria and fungal spores that are always around.   Immune systems were formed over millions of years of evolution in response to the regular hazards over that period and modern societies have been around for a blink of an eye in that context.  The immune systems would need some help with new challenges although that will eventually fix itself.
 
On the other hand we do need to look at whether immunizations do something like pushing down one side of a seesaw.

No leadership in Oakland

 
Check the above link for an example of how our traditional systems aren’t working.   The bay bridge in Oakland has spans that fall down in the 89 earthquake and they have the option of fixing it cheap and quickly and instead they embark on some boondoggle at enormous expense, with the intention of replacing the bridge with something less safe than the initial structure at prohibitive cost.  
 
Do I smell a bit of Illinois here? How connected are the contractors that must be involved?   I see one beneficiary of the boondoggle and it ain’t the people. 
 
You could spend a trillion dollars on Oakland and the place would still be ugly.   Trying to jazz it up a bit with some kind of world class landmark isn’t going to change much other than wasting a lot of public money.  
 
The cost in time and money of putting up a bridge and the importance of bridges for travel efficiency militate in favor of having engineers make all of the important decisions and then if some parasites that want to make more money generate some artificial "groundswell" of support for a grandiose option it is the job of the politicians to band together across party lines and stomp on it.
 
Some populism can have an important function where systems become out of touch with the needs of society but populism can have a dark side if it degenerates into rubber stamping ideas that have become popular through incomplete or inaccurate information.    Part of leadership is about taking on positions that are unpopular but necessary and taking the steps to make the right decision popular.