The Predictive Value of Orofacial Esthetic Components on Quality of Life: Pilot Insights from a Serbian Sample using Adaptive Neuro-Fuzzy Inference System.

Abstract

Aim: To evaluate orofacial esthetic satisfaction using the Serbian version of the Orofacial Esthetic Scale (OES-SERB) and to explore whether a hybrid artificial intelligence model can identify the esthetic domains associated with self-reported quality of life.

Methods: This pilot study included 63 adult participants (aged 20-50) who completed the OES-SERB questionnaire following clinical dental examination. The adaptive neuro-fuzzy inference system (ANFIS), which combines neural networks and fuzzy logic, was applied to assess the predictive relationship between seven OES-SERB domains and a global self-rated quality-of-life score. The data set was divided into a training subset (70%) for model development and a validation subset (30%) for internal validation.

Results: Participants without orthodontic anomalies or tooth loss reported significantly higher satisfaction across most OES-SERB domains (P = 0.001 to 0.032). The ANFIS model demonstrated limited generalization, with high validation error across predictors. Among all domains, satisfaction with the shape/form of teeth showed the most stable predictive performance, whereas gingival appearance did not generalize well due to overfitting.

Conclusion: The OES-SERB is a promising instrument for evaluating orofacial esthetic perceptions in the Serbian population. Although the ANFIS-based analysis provided exploratory insights into potential predictors of perceived quality of life, the model's restricted generalization underscores the need for studies with larger samples and alternative statistical approaches to confirm these preliminary observations.