Data analysis with Shapley values for automatic subject selection in Alzheimer’s disease data sets using interpretable machine learning

Table 6 Mean RF accuracies and F1-scores (in %) for the independent ADNI test set (no information rate 55.56%)

Exclusion method	Zero-cutoff			Max valid			Max test
	n	ACC (in %)	F1 (in %)	n	ACC (in %)	F1 (in %)	n	ACC (in %)	F1 (in %)
Random (-)	–	–	–	41	63.47	66.49	449	67.01	73.77
LOO (LR)	84	63.06	65.75	102	64.31	66.70	5	64.44	68.23
LOO (RF)	49	63.47	66.99	23	63.33	67.09	273	65.42	68.50
Data Shapley (LR)	152	65.76	68.78	340	65.49	67.18	329	68.89	70.33
Data Shapley (RF)	164	66.88	72.21	134	66.25	70.96	321	68.82	74.53

Different methods were used to identify and focus on the training subjects with the most informative data. The zero-cutoff method excluded all training subjects with Data Shapley values smaller than zero. Max valid was the threshold achieved by maximizing the results for the independent validation set. Max test was the optimistic threshold which achieved the best results for the test set. Ten repetitions with different seeds were performed for every exclusion data set. The best results are highlighted in bold

ISSN: 1758-9193