Skip to main content
. Author manuscript; available in PMC: 2021 Nov 16.
Published in final edited form as: JID Innov. 2021 Sep 2;1(3):100037. doi: 10.1016/j.xjidi.2021.100037

Table 7.

Representative1 combinations for parameters selected by LASSO regression as the most important variables in differentiating sclerotic cGVHD patients from post-HCT controls for each site.

Frequency Relaxation Time Stiffness Creep Decrement AUC with overfit correction [95% CI]
Abdomen L + + + + 0.884 [0.674, 0.996]
Abdomen R + + + 0.902 [0.737, 0.999]
Upper Arm L + + 0.773 [0.431, 0.981]
Upper Arm R + + 0.773 [0.422, 0.983]
Chest L + + + + 0.801 [0.437, 0.992]
Chest R + 0.829 [0.546, 0.992]
Upper Back L + + 0.840 [0.533, 0.994]
Upper Back R + + 0.756 [0.358, 0.987]
Shin L + + 0.855 [0.606, 0.992]
Shin R + + + 0.761 [0.407, 0.978]
Calf L + 0.741 [0.404, 0.964]
Calf R + + + + + 0.780 [0.411, 0.986
Lower Back L + + + 0.812 [0.445, 0.988]
Lower Back R + + + + 0.740 [0.327, 0.979]
Shoulder L + + + + + 0.725 [0.381, 0.955]
Shoulder R + + + + + 0.757 [0.392, 0.981]
Dorsal Forearm L + + 0.782 [0.427, 0.985]
Dorsal Forearm R + + 0.846 [0.559, 0.995]
Average across all sites + + 0.871 [0.641, 0.998]
Average across supine sites + + 0.778 [0.429, 0.978]
1

Representative: out of 10 repetitions, the most commonly selected combination of parameters for each site is shown.