Table 4.

Performance of ML (accuracy, sensitivity, specificity, GM, and Dominance) in the classification of CDR = 0.5 versus CDR = 0 using linear, quadratic, Gaussian RBF, and Multilayer Perceptron kernels. Results were obtained using the computation-based feature reduction.

Kernel	Accuracy [mean ± std]∗	Sensitivity [mean ± std]∗	Specificity [mean ± std]∗	Geometric Mean [mean ± std]∗	Dominance [mean ± std]∗
Linear	0.86 ± 0.07	0.85 ± 0.10	0.87 ± 0.10	0.86 ± 0.07	−0.01 ± 0.15
Quadratic	0.86 ± 0.07	0.85 ± 0.11	0.88 ± 0.09	0.86 ± 0.07	−0.03 ± 0.15
Gaussian RBF	0.86 ± 0.07	0.85 ± 0.10	0.87 ± 0.10	0.86 ± 0.07	−0.02 ± 0.15
Multilayer Perceptron	0.85 ± 0.07	0.83 ± 0.12	0.87 ± 0.10	0.85 ± 0.07	−0.04 ± 0.16

^∗Averaged across 10 rounds of the nested CV and across 100 iterations.