Table 2.

Accuracies of ten trials of three-fold cross-validation in prostate. Validation and test accuracies for a single slide video of prostate adenocarcinoma (patient 1, slide at upper right in Fig 2, performance plotted at right in Fig 9)), left side of the slide versus right side.

Testing the best classifier on draws of the data on the second prostate patient, accuracies are 0.944, 1, 0.944, 0.944, 1, 0.944, 0.944, 0.944, 1, and 1.

Direction	Trial	Fold0 Valid	Fold1 Valid	Fold2 Valid	Valid Acc	Fold0 Test	Fold1 Test	Fold2 Test	Test Acc
leftright	0	0.9992	0.9946	1.0000	0.9979	0.7778	0.7778	0.7778	0.7778
leftright	1	0.9512	0.7282	0.9994	0.8929	0.8889	0.8889	0.8889	0.8889
leftright	2	0.9550	1.0000	0.6530	0.8693	0.8889	0.7778	0.7222	0.7778
leftright	3	0.8636	0.9992	1.0000	0.9543	0.8889	0.7778	0.9444	0.9444
leftright	4	0.8276	0.7760	0.9940	0.8659	0.8889	0.7778	0.8889	0.8889
leftright	5	0.8654	0.9986	1.0000	0.9547	0.9444	0.9444	0.9444	0.9444
leftright	6	0.8560	0.9862	0.9992	0.9471	0.8889	0.8889	0.8889	0.8889
leftright	7	0.8674	1.0000	0.9984	0.9553	0.8889	0.8333	0.8889	0.8333
leftright	8	0.9560	0.8560	0.8760	0.8960	0.8889	0.8333	0.9444	0.8889
leftright	9	0.6846	0.9992	0.9560	0.8799	0.8333	0.8333	1.0000	0.8333
rightleft	0	0.9786	0.7760	0.9146	0.8897	0.8889	0.9444	1.0000	0.8889
rightleft	1	1.0000	0.7292	0.8460	0.8584	1.0000	0.9444	1.0000	1.0000
rightleft	2	0.7130	0.9512	0.8676	0.8439	0.8889	1.0000	0.8333	1.0000
rightleft	3	0.9998	1.0000	0.9664	0.9887	0.9444	0.9444	1.0000	0.9444
rightleft	4	0.7760	1.0000	0.8842	0.8867	0.8889	0.9444	1.0000	0.9444
rightleft	5	0.9758	0.9984	0.5926	0.8556	0.9444	0.8889	0.9444	0.8889
rightleft	6	0.6344	0.9770	1.0000	0.8705	0.8889	1.0000	1.0000	1.0000
rightleft	7	0.7760	0.9028	1.0000	0.8929	0.8889	1.0000	1.0000	1.0000
rightleft	8	0.8560	0.8560	0.9992	0.9037	0.9444	0.9444	0.9444	0.9444
rightleft	9	0.8560	0.9412	0.8538	0.8837	1.0000	1.0000	1.0000	1.0000