Prognostic models examples

External validation of a model means determining whether it performs well in groups of patients other than those on whom it was derived that is, how is it likely to do in real clinical practice. These other groups almost certainly will differ in case mix, referral patterns, treatment protocols, methods of measurement of variables and definition of outcomes. Nevertheless, if a prognostic model includes powerful predictive variables, appropriately modeled, it should vaUdate reasonably well in other groups of patients. For example. Fig. 14.1 shows the vaUdation of the ABCD score on pooled individual patient data from six independent groups of patients with TIA (Johnston et al. 2007) (Ch. 15). 

Combination of elements of the above methods (hybrid methods). For example, the output of a prognostic model can be combined with observed data as an input to a diagnostic model. 

Exposure assessment is done under the strong assumptions that (1) an adequate model for exposure calculation is on hand and (2) sufficient data about all influential exposure factors are available. The calculation is a prognosis about the expected level of exposure or the burden. Direct methods of exposure assessment, such as personal sampling (air, radiation), duplicate studies (nutrition) and human biomonitoring, provide information on a measurement level. The exposure assessors and the risk managers should balance the reasons for using prognostic techniques instead of direct exposure measurement methods. Both should anticipate critical questions about the validity of the exposure assessment technique in the course of public risk communication. Questions heard by the authors from concerned persons include, for example ... 

The spatial resolution of a model in the vertical direction is normally designated by the number (integer L) of vertical levels or layers necessary to cover from the earth s surface to the model top. Thus, L42 means that the model has 42 vertical layers. In order to estimate the computer speed necessary to perform a 10-day medium-range forecast, let us take a T170L42 model as an example. For this model the mrmber of grid points that covers the entire globe and depth of the atmosphere is about 2.4 million (=340 X 170 X 42). The ntunber of prognostic variables is foirr (i.e., u, v, T, and q) for the hydrostatic model. Thus, the total mrmber of discretized dependent variables is about 10 million. Now, a typical time step for this model would be 5 min. This means that nearly 3000 repeated calculations are needed to make one 10-day forecast (=10 d X 24 h X 60 min/5 min). 

Moons, K.G.M., Donders, A.R.T., Steyerberg, E.W. Harrell, FE. 2004 Penalized maximum likelihood estimation to directly adjust diagnostic and prognostic prediction models for overoptimism a clinical example. Journal of Clinical Epidemiology, 57, 1262-1270. 

---