Evaluation of success criteria in postpeak monoblock applications with artificial intelligence method

Objective: The aims of this retrospective study were to compare changes in Class II monoblock treatment applied in the Peak and Postpeak periods of growth development by evaluating cephalometric parameters and cooperation factors, and to determine critical decision points in evaluating treatment success with artificial intelligence analysis methods. Method: In this study, patients who were treated with a monoblock appliance between the years 2017 and 2022 were examined. Radiographs of 74 Postpeak period patients (MP3u ve Ru) with an mean age of 15±1.37 years and 120 Peak period (MP3cap) patients with an mean age of 12.8±1.35 years were included in the study. Lateral cephalometric measurement data taken which were obtained before and after monoblock treatment were analyzed with Recursive Partitioning Analysis (RPA) in decision tree analysis. Recursive Partitioning and Regression Trees (RPART) function in the R programming language was used in the analysis. Classification and Regression Trees (CART) method was used in Recursive Partitioning Analysis. Results: Although decreases in ANBº and Wits measurements and increases in mandibular linear measurements (Co-Gn, Co-Go, Go-Me) were observed in both study groups, these effects were significantly higher in Peak period than in Postpeak period (p<0,001). However, Overjet elimination and profile correction were similar between the two study groups, lower incisor protrusion was significantly higher in Postpeak period. When cooperation increased, the restrictive effect on the maxilla and the improvement in maxillo-mandibular measurements increased. As a result of the decision tree analysis, critical decision points were determined in the skeletal parameters (Pg-xTot ≥ 2.4 mm; Co-Gn ≥ 3.8 mm and 5.9 mm) and the dental parameters (A1-NBº < 8.2º and 8.1º; Ü1-NAº ≥ -0.75º; Overbite ≥ -1.5 mm). Conclusion: With monoblock treatment in Postpeak period, 38% skeletal and 68% dental effect were observed. Decision trees showed 88% and 85% accuracy. The treshold volves from the decision tree analysis provided both skeletal and dental critical decision points in evaluating the success of functional orthopedic treatment in the Postpeak period