Table 5

Comparison of predictive discriminatory accuracy of cross-validated models using measures of activity versus attention
Criteria Linear Discriminant Analysis General Linear Model Multinomial Logistic Regession
No 4's Attn CPT-II Attn Activity No 4's Attn CPT-II Attn Activity No 4's Attn CPT-II Attn Activity
Accuracy 0.659 0.670 0.768ab 0.650 0.663 0.766ab 0.703 0.706 0.801ab
Kappa 0.257 0.281 0.505ab 0.240 0.269 0.504ab 0.360 0.360 0.583ab
ROC-AUC 0.671 0.712 0.821ab 0.661 0.692 0.800ab 0.722 0.729 0.839ab
Sensitivity 0.786 0.790 0.835cd 0.752 0.776 0.816ef 0.800 0.821 0.829
Specificity 0.462 0.482 0.665ab 0.491 0.486 0.687ab 0.552 0.528 0.758ab
Criteria Neural Network Support Vector Machine Random Forest Classification
No 4's Attn CPT-II Attn Activity No 4's Attn CPT-II Attn Activity No 4's Attn CPT-II Attn Activity
Accuracy 0.659 0.677 0.809ab 0.720 0.709 0.817ab 0.663 0.624 0.842ab
Kappa 0.273 0.299 0.607ab 0.396 0.376 0.621ab 0.292 0.191 0.676ab
ROC-AUC 0.687 0.711 0.816ab 0.731 0.753 0.829ab 0.661 0.673 0.889ab
Sensitivity 0.742 0.769 0.801ab 0.811 0.793 0.807 0.753 0.775 0.828ab
Specificity 0.528 0.534 0.802ab 0.578 0.578 0.833ab 0.543 0.415 0.870ab

aActivity > No4's Attention p < 10–15, bActivity > CPT-II Attention p < 10–15.

cActivity > No4's Attention p < 10–6, dActivity > CPT-II Attention p < 10–5.

eActivity > No4's Attention p < 10–8, fActivity > CPT-II Attention p < 0.0001.

Teicher et al.

Teicher et al. BMC Psychiatry 2012 12:190   doi:10.1186/1471-244X-12-190

Open Data