Thanks for those additional quotes. And it's worth remembering that you'd get a 50% classifier rate by tossing a coin!Thanks OceamBlue - Yes, I completely missed that.. how I did I don't know. How unfortunate that was. Gene expression has been problematic to say the least in ME/CFS...Here's another section from the paper:
a model is typically trained by maximizing its performance on some set of training data. However, its efficacy is determined not by its performance on the training data but by its ability to perform well on unseen data.
Thanks for those additional quotes. And it's worth remembering that you'd get a 50% classifier rate by tossing a coin!
I think failure to replicate the initial findings on an independent sample is a common failing for CFS research, particularly where biomarkers are postulated. With a single sample, thresholds are set to maximise the preidctive power for that particular set of patients and controls, which is 'overfitting' data. This from Wikipedia explains the issue:
Haven't we been thru this with florinef? Can anybody raise their hand and say "this drug helped my ME/CFS symptoms"? I mean on a long term basis? I went thru all the tilt table stuff and actually spoke with Dr. Rowe afterwards. He is a consummate gentleman, a first rate researcher and medical practioner, and he really cares about ME/CFS patients; but this theory that taking IO negating drugs are helpful, in ME/CFS, is totally wrong.
As Mitt Romney says, I'll bet $10,000, or one of my wife's two Cadillacs, or or.....
AbstractClin Auton Res. 2011 Dec;21(6):419-23. doi: 10.1007/s10286-011-0130-x. Epub 2011 Jul 28.
Fludrocortisone improves nausea in children with orthostatic intolerance (OI).
Fortunato JE, Shaltout HA, Larkin MM, Rowe PC, Diz DI, Koch KL.
Source
Department of Pediatrics, Division of Pediatric Gastroenterology and Nutrition, Wake Forest University School of Medicine, Winston-Salem, NC, USA. jfortuna@wfubmc.edu