Figure 4 | Scientific Reports

Figure 4

From: Risk of data leakage in estimating the diagnostic performance of a deep-learning-based computer-aided system for psychiatric disorders

Figure 4

Mean and standard deviations of classification accuracies of four cross-validation (CV) strategies for two deep-learning models with respect to the different window lengths. For both deep-learning models, the classification accuracies of tCV and otCV are significantly higher than those of sCV for all window lengths, and those of otCV do than those of tCV, except the 60-s window length (original data) (Bonferroni corrected p < 0.05: otCV > tCV > sCV). In terms of osCV, the deeper model of CNN-13 shows improved classification performances compared to sCV, whereas the shallower model of EEGNet shows little changes in classification accuracy. n.s. means no significance, and the pairs without n.s. showed significant difference in terms of the classification performance.

Back to article page