Radial Basis Function-Sparse Partial Least Squares for Application to Brain Imaging Data

Table 2

The result for sPLS without basis expansion and with, respectively, for 100 simulated data sets.

Knots distance	𝑃	𝑛	1st component			2nd component
			Sensitivity	Specificity	C-index	Sensitivity	Specificity	C-index
Original sPLS: without basis expansion
0	40	50	0.26	0.99	0.25	0.30	0.99	0.29
		100	0.34	0.99	0.33	0.39	0.99	0.38
	80	50	0.37	0.99	0.36	0.43	1.00	0.43
		100	0.39	0.99	0.38	0.44	1.00	0.44
RBF-sPLS: with basis expansion
2	40	50	1.00	0.60	0.60	1.00	0.68	0.68
		100	1.00	0.86	0.86	1.00	0.87	0.87
	80	50	1.00	0.73	0.73	1.00	0.75	0.75
		100	1.00	0.84	0.84	1.00	0.88	0.88
4	40	50	1.00	0.29	0.29	1.00	0.13	0.13
		100	1.00	0.29	0.29	1.00	0.04	0.04
	80	50	1.00	0.27	0.27	1.00	0.13	0.13
		100	1.00	0.21	0.21	1.00	0.08	0.08
8	40	50	1.00	0.08	0.08	1.00	0.06	0.06
		100	1.00	0.05	0.05	1.00	0.00	0.00
	80	50	1.00	0.05	0.05	1.00	0.01	0.01
		100	1.00	0.02	0.02	1.00	0.00	0.00