![\"20140205-091454.jpg\"](\"https:\/\/stg-blogs.bmj.com\/adc\/files\/2014\/02\/20140205-091454.jpg\")
There are a variety of clinical prediction rules in the world. If\u00a0you’ve seen one –\u00a0they always used to have a nomogram attached – it would take the answers to a few questions and come up with a ‘probability of bad thing happening’.<\/p>\nThere are a variety of clinical prediction rules in the world. If\u00a0you’ve seen one –\u00a0they always used to have a nomogram attached – it would take the answers to a few questions and come up with a ‘probability of bad thing happening’.<\/p>\n
As we’ve mentioned previously<\/a>, there’s an issue with deriving models\u00a0and assuming they will\u00a0work as well in every other place you try them (and have a read here <\/a>and here <\/a>if you want a fuller explanation\u00a0from the E&P section). A new model needs to be ‘validated’ – that is, checked to see if it works.<\/p>\n Two bits could\u00a0be looked at:<\/p>\n a) Calibration – does the ‘predicted probability’ from the model match(~ish) the observed probability in the new set<\/p>\n b) Discrimination – if you set a threshold (say, less than 10% chance = low risk), does the model have the same sort of sensitivity and specificity<\/a> as the original data?<\/p>\n How close<\/a> you need these to be to claim effective validation is, as many things, a matter of clinical judgement<\/a> but it’s worth understanding what they are validating to make a sensible assessment.<\/p>\n – Archi<\/p>\n","protected":false},"excerpt":{"rendered":" There are a variety of clinical prediction rules in the world. If\u00a0you’ve seen one –\u00a0they always used to have a nomogram attached – it would take the answers to a few questions and come up with a ‘probability of bad thing happening’. As we’ve mentioned previously, there’s an issue with deriving models\u00a0and assuming they will\u00a0work […]<\/p>\n