RESUMEN

BACKGROUND: As artificial intelligence within digital processes continues to advance and replace conventional manual workflows, it is crucial that digital data are consistent with analog data. The aim was to evaluate the validity and time efficiency of digital cast analysis on digital models in comparison with the manual, gold standard, cast analysis on plaster models. METHODS: Cast analysis was performed on 30 patients in three various methods: manually measured variables on plaster models (MP), manually measured variables on digital three-dimensional models (MD), and automatically measured variables on digital three-dimensional models (AD) on digital models. Digital cast analysis was performed in CS Model+. Analyses included metrical and categorical variables and the required work time. Measurements in MD and AD were validated to MP. Validity of the metrical variables was analyzed with Bland-Altman, Dahlberg's formula, and paired sample t test. Categorical variables were validated by Cohen's Kappa. Work time was analyzed with Wilcoxon signed-rank test. RESULTS: Metrical variables had measurement errors ranging 0.4 to 1.4 mm between MP-MD, and 0.6 to 3.2 mm between MP-AD. Observations of categorical variables had a moderate to strong (0.65 to 0.9) level of agreement between MP-MD, and a weak to moderate (0.4 to 0.68) level of agreement between MP-AD. Data for dental stage, vertical, and transversal relation was not provided in AD. Cast analysis was performed quicker digitally, P ≤ 0.05. CONCLUSIONS: Digital cast analysis is consistent with manual cast analysis for metrical variables. Analyses of categorical variables show a weak level of agreement with automatic digital analysis, such as space conditions and midline assessments. Digital cast analysis optimizes time compared with manual cast analysis, with automatic analysis being the fastest.