Assessment of Perceived Facial Age Changes Following Orthognathic Surgery Using Artificial Intelligence

Orthognathic surgery is performed to improve facial esthetics and function in patients with dentofacial deformities. This study aimed to evaluate the impact of orthognathic surgery on perceived facial age using an artificial intelligence (AI)-based age-estimation tool and to examine accompanying changes in self-rated flourishing and social appearance anxiety.

This retrospective, single-arm, before-and-after study included 27 patients who underwent bilateral sagittal split ramus osteotomy (BSSRO) and/or Le Fort I osteotomy. Preoperative and postoperative standardized frontal photographs were analyzed with the AI-based application "AgeBot: How Old Do I Look?". Participants also completed the Social Appearance Anxiety Scale (SAAS) and the Flourishing Scale (FS; Turkish adaptation by Telef). Both questionnaires were completed at a single postoperative interview-first for the remembered preoperative state and then for the current postoperative state-so the psychosocial pre-post comparisons reflect retrospectively perceived change. Normality was assessed on the paired differences; the primary AI-age outcome was analyzed with the paired t-test (confirmed by the Wilcoxon signed-rank test) and the psychosocial outcomes with the Wilcoxon signed-rank test, with effect sizes reported. A data-informed sensitivity analysis, rather than an a priori power calculation, accompanied the primary outcome.

AI-estimated facial age did not change significantly after surgery (mean difference [preoperative-postoperative] 0.11 ± 2.49 years, 95% CI -0.87 to 1.09, p = 0.818; Cohen's dz = 0.04), and the confidence interval excluded the previously reported 1.31-year reduction. Exploratory change-score comparisons showed no significant differences by sex or smoking status. Postoperative SAAS scores were lower than retrospective preoperative scores (median 37 [IQR 24.5-42.0] vs. 51 [43.0-56.0]; Z = -4.07, p < 0.001, r = 0.55), and postoperative FS scores were higher (median 46 [40.0-51.5] vs. 45 [38.5-45.0]; Z = -2.45, p = 0.014, r = 0.37).

Orthognathic surgery was not associated with a statistically significant change in AI-estimated facial age in this small cohort. More favorable social appearance anxiety and flourishing scores were observed following treatment, but the retrospective, uncontrolled assessment precludes causal attribution. AI-assisted facial age estimation may complement patient-reported outcomes, but validated, repeatability-tested tools are required before it can serve as a primary outcome measure.