Extended Data Fig. 10. Neural fragility correlation against non-epileptic clinical covariates.

Neural fragility correlation against non-epileptic clinical covariates - Fragility success probabilities (denoted as “Confidence Statistic” in y-axes) split by clinical factors, such as handedness (a), gender (b), ethnicity (c) and age at surgery (d). Not all patients had data for each of these categories, so the subset of available data was used. Note the sample sizes vary across different groups shown. For ethnicity, we also had 1 Asian subject, but left it out because the permutation effect size estimation procedure does not work for 1 sample. Effect sizes were estimated using the permutation test and Mann Whitney U test described in Methods. The corresponding effect sizes and p-values were (0.1/0.99) for handedness, and (0.12/0.7) for gender. The pvalue was computed using the one-sided Mann-Whitney U test. The slope was negligibly close to 0 for surgery age linear fit. There was no relatively significant trend in the data related to ethnicity. The significant Cohen’s D effect size difference is primarily due to the low sample sizes in non-Caucasian ethnicities. The error bars represent 95% confidence interval specified by 2 standard deviations.