R statistical package version

The sample size was calculated using the R statistical package, version 3.3.1 (20 December 2018, © 2018, R Foundation for Statistical Computing, Vienna, Austria) [14 ]. Cochran’s sample size formula for prevalence studies was used to determine the proper sample size [15 ]. The expected proportion of participants who consulted an orthodontist for their perceived dental problems was determined according to the study of Alnaafa et al. [16 ], assuming a proportion of 72.6%. A more conservative expected proportion of 30% was considered in the present study.
At a 95% confidence interval and 2% margin of error, the estimated sample size was 2017 participants. An additional 50% contingency for non-response increased the sample size to 3000 participants.

The intraobserver (two consecutive measurements by HN) and interobserver (measured by HN and JCH) reproducibility of measurements of the OCT parameters were assessed by calculating the intraclass correlation coefficients (ICCs). SD-OCT images of 20 randomly selected patients were used for this analysis. To compare the mean values of continuous variables between the two subgroups, the Mann–Whitney U test was used. A Chi-square test or Fisher’s exact test was performed for comparison of categorical variables. Correlations between the parameters and AL and correlations among the parameters were evaluated by Pearson’s correlation analysis, and the correlation coefficient (R) was calculated. The Brown–Forsythe test was performed to assess the equality of distribution of the parameters. An AL of 25 mm was arbitrarily set as a cutoff for comparison between the two subgroups according to AL. A logistic regression analysis was performed to confirm the factors associated with the location of the dominant VF defect. A p-value < 0.05 was considered statistically significant. Statistical analyses were performed using IBM SPSS software version 24.0 (IBM Corp., Armonk, NY, USA) and R statistical package version 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria).

The continuous variables were analyzed using the independent t-test or Wilcoxon rank-sum test, and the categorical variables were analyzed and compared by the Chi-squared or Fisher’s exact tests. Unadjusted odds ratios with 95% confidence intervals (CIs) for each variable of the study outcome in the training cohort were calculated using the logistic regression analysis. All statistical analyses were conducted with SAS software version 9.4 (SAS Institute, Cary, NC, USA) and R Statistical Package version 3.4.3 (www.R-project.org (accessed on 1 October 2022)). Furthermore, we used TensorFlow version 1.13.1 (www.tensorflow.org (accessed on 1 October 2022)) and scikit-learn version 1.0.2 (www.scikit-learn.org (accessed on 1 October 2022)) libraries in the python (version 3.6.9, www.python.org (accessed on 1 October 2022)) programming environment for machine learning modeling. The statistical significance criterion was set to be two-sided, and p values < 0.05 were considered statistically significant.