So neuros are invited to learn about statistics. So whilst they are off to Naples to learn, as you tuck into your frozen pizza (it’s almost Naples after all) you too can learn some statistics which will help to determine how much you log-in to your memory banks that will help you to read science papers. I bet this won’t be presented in Naples
There is one thing you and the neuros need to understand about is that when you read about MRI and Biomarkerologists that claim to PREDICT the future…Statistically speaking….you need to learn that in many clinical papers.The r = correlation coefficient value is not P for prediction but actually P for “pants”. When I see r=0.98 it is like a thick juicy steak (Meat of Jack fruit or whatever floats your boat) that gets my attention I will even get interested when r=0.8 but when it is r=0.4…Sadly in my head I see a pile of undies (I know that pants in the US are typically considered to be called trousers but pants comes from underpants meaning shreddies, knickers, drawers, skivvies etc).
An r value of 1 is a perfect correlation meaning as one goes up the other variable goes up also. An r=-1 is a perfect negative correlation, so as one goes up the other outcome goes down. r=0 means no correlation.
So often in the P for prediction they do correlations of relating this MRI or that biomarker measure to clinical outcome. What you need to know is that here, they are talking about predictions of a general population (so what trend will happen) so an r=0.2-0.6 means that if one variable goes up the other goes up so more neurofilament levels (mark of nerve breakdown) in blood means it is more likely you will have clinical progression in the future. This is where the P for prediction comes from however when r=0.2 the P for prediction of events occuring in the individual is not P for prediction but P for pants of what happens. Because of this many imaging outcome never translate into clinical practice.
disclaimer my views only coi None
Source: multiple-sclerosis-research.org