Skip to main content
. 2021 Jun 14;22(12):6345. doi: 10.3390/ijms22126345

Figure 4.

Figure 4

Potential for the most favored model. (a) Decision curve analysis of models for prediction of phenotypic outcome (LFS or HBC). The y-axis indicates the net benefit of using mod_3v model (red line). The thin gray line (All) shows net benefit values expected assuming all assessed variants are LFS. The darker gray line (None) shows net benefit values expected assuming no assessed variants are LFS. The net benefit for prediction of LFS variants is regarded as positive for probability values exceeding those for the “All” and “None” values. (b) The nomogram that facilitates manual estimation of the risk of LFS disease outcome using the mod_3v model. For the value of each explanatory variable the equivalent value on the “Points” scale is assessed. The sum of all Points values is then located on the “Total Points” scale (middle green row) so that the corresponding probability value can be read from the “LFS outcome rate” scale (upper green row). The dotted arrows show a hypothetical example for a surface residue with a comp_dif value of 50 and a PPI6_dif value of −500, for which the equivalent “Point” values (10, 40 and 75) summate to 125 (“Total Points”), giving a LFS outcome risk of slightly over 0.3.