Originally posted by rohrer01
View Post
You can fudge the figures in any study. Just throw out the few bits of data that don't fit the curve and viola! Just how many "bits" are getting thrown out? How many are allowed to be thrown?
I certainly hope they have at least one and hopefully two mathematical biologists and at least one statistician review that work. I don't know exactly what they are doing but it seems like they found a few hundred markers that occurred in various combinations in various patients with differing curve trajectories. They then took various combinations (300C50 for example) using massive computing assets to fit various perhaps even weighted combinations of these markers until they empirically got the best fit to the observed data. So it ended up being a 53 variable equation as the best fit. They then had it kick out a number between 0 and 200 for various regions of this curve. But I think a potential issue is whether that type of equation could eve possibly yield a unique solution if that is what they are doing.
I don't know what they did but that's my guess.
Let's see if they ever publish. Maybe Pnuttro will be right and it won't reach the printed page. Just reaching the press conference stage has a long history of going nowhere.
Comment