Smile evolution in the social media era: a longitudinal analysis of Miss Italia winners (1982–2025)

Introduction

Smile aesthetics represent a fundamental component of dentofacial harmony and are central to contemporary orthodontic and prosthodontic planning [1–13,15,18,20]. Over the past decades, increasing emphasis has been placed on soft-tissue dynamics, lip mobility, and dental display as determinants of an attractive smile [9–11,15,18,20]. From a clinical standpoint, the smile is not merely a static configuration but a dynamic interplay of muscular activity, dental exposure, and gingival architecture [1–2,8,11]. Variables such as upper incisor display, smile arc, buccal corridors, and lip curvature have been extensively studied in relation to perceived attractiveness [4,9,13,18,20]. Longitudinal and morphometric studies of facial attractiveness further support the integration of dental, soft-tissue, and facial parameters in aesthetic assessment [14,16–17,19].

In parallel with clinical developments, the increasing dominance of digital media has reshaped aesthetic standards [24,25]. Social media platforms prioritize visual immediacy and emotional expressivity, potentially influencing how smiles are performed and perceived. In such contexts, highly visible smiles may be favored because of their communicative efficiency and photographic recognizability.

The identification of 2004 as the dividing line between the pre-social and social media eras serves as a pragmatic chronological marker rather than a biological threshold. This year is used as a conceptual proxy for the beginning of a more image-driven digital environment, in which facial appearance increasingly became part of everyday social interaction. Before this period, digital imagery often functioned as a static archive; after the emergence of social platforms and photo-sharing environments, images became a primary means of social communication. Consequently, the smile may increasingly be interpreted as a visual communication tool optimized for the digital age. The Miss Italia pageant provides a unique longitudinal framework through which such changes can be observed. This study aims to quantitatively evaluate the evolution of smile characteristics across pre- and post-social media eras, integrating FACS-based metrics with AI-assisted analysis [10,21,23–25].

Materials and Methods

This retrospective observational study included 44 Miss Italia winners from 1982 to 2025. Subjects were divided into two equal groups: the pre-social-media era (1982–2003) and the social-media era (2004–2025). Smile analysis was conducted using a standardized grid derived from the Facial Action Coding System (FACS) [21]. The evaluated parameters included AU12, AU6, AU25, AU26, dental exposure, symmetry, curvature, and intercommissural width [8–11,13,15,18,20–21]. These variables are consistent with established determinants of smile aesthetics in dental literature.

A key methodological innovation of this study is the use of artificial intelligence-assisted facial analysis to evaluate smiles. All images were processed using an AI system specifically trained for qualitative FACS-based scoring [21,23]. The model was calibrated using standardized coding protocols and validated through internal consistency checks. AI-assisted facial analysis has been increasingly adopted in research settings for its reproducibility and ability to minimize observer-related variability [23,26]. Nevertheless, the AI output was interpreted as a standardized image-based assessment rather than a substitute for clinical smile evaluation.

Two composite indices were computed: the Simple Smile Index and the Weighted Smile Index, the latter emphasizing AU12, AU6, and dental exposure [8,10,11,20]. Statistical analysis included Welch’s t-test, Mann-Whitney U test, Cohen’s d effect size estimate, Spearman’s correlation, and regression modeling with robust standard errors [27–28]. Interrupted time-series models were used to assess structural changes over time.

Statistical Analysis

A comprehensive statistical framework was adopted to ensure robust comparison between the two temporal cohorts and to explore both group differences and longitudinal trends. Given the relatively small sample size and the potential for unequal variances between groups, Welch’s t-test was selected as the primary method for comparing mean values of continuous variables, including the Simple Smile Index, Weighted Smile Index, Dental Exposure, and AU12 activation. Unlike the standard Student’s t-test, Welch’s test does not assume homoscedasticity and is therefore appropriate when variance equality cannot be guaranteed. The results consistently demonstrated highly significant differences between the two periods (p <= 0.0004), indicating that the observed increases in smile-related metrics were unlikely to be due to random variation.

To further strengthen the reliability of these findings, a non-parametric Mann-Whitney U test was conducted as a confirmatory analysis. This test does not rely on assumptions of normal distribution and is particularly suitable for ordinal or non-normally distributed data, such as FACS-based scores derived from qualitative assessment. The concordance between parametric and non-parametric results reinforced the robustness of the observed differences and reduced the likelihood of statistical artifacts.

In addition to statistical significance, Cohen’s d was estimated to assess the magnitude of differences between groups [27]. The resulting values, ranging from 1.25 to 1.51, indicate large effect sizes, suggesting that the differences are not only statistically significant but also clinically and practically meaningful. From a clinical interpretation standpoint, the observed increase in dental exposure, with a mean difference of approximately +1.27 units, can be interpreted as a noticeable augmentation of maxillary incisor display during posed smiling, a parameter that, in the orthodontic and prosthodontic literature, is typically associated with enhanced smile attractiveness and perceived youthfulness [1–2,9,18,20].

To investigate temporal dynamics beyond binary group comparisons, Spearman’s rank correlation coefficient was used to assess associations between each variable and the chronological year. This non-parametric measure was chosen for its robustness to non-linear relationships and outliers. Strong positive correlations were identified for both composite indices and dental exposure (rho = 0.69), indicating a consistent upward trend over time. Moderate but significant correlations were also observed for AU12 (rho = 0.48) and AU6 (rho = 0.34), suggesting progressive intensification of muscular activation involved in smiling.

The relationship between categorical variables, specifically the distribution of Duchenne versus non-Duchenne smiles, was evaluated using the chi-square test of independence [22]. No statistically significant association was found between smile type and temporal period (p = 0.469), indicating that while smile intensity increased, the qualitative classification of smiles remained relatively stable across decades.

Finally, to model temporal trends while accounting for potential structural changes associated with the emergence of social media, linear regression models with robust (HC3) standard errors were implemented [28]. Both additive models and interrupted time-series models were tested. Although regression coefficients generally supported a positive temporal trend, statistical significance was limited in some models, likely due to sample size constraints and variability in the data. Notably, dental exposure remained significantly associated with time (p = 0.030), reinforcing its role as a key evolving parameter. Overall, the combined use of parametric, non-parametric, and regression-based approaches provides a comprehensive assessment of the data.

Results

The analysis demonstrated a consistent increase in smile expressivity over time. Mean values for the Simple Smile Index increased from 11.36 in the pre-2004 group to 13.73 in the 2004–2025 group, while the Weighted Smile Index rose from 15.91 to 19.59. Dental exposure increased markedly, from 2.45 to 3.73, accompanied by a significant rise in AU12 activation. These differences were statistically significant across all primary variables (p <= 0.0004), with large effect sizes (Cohen’s d ranging from 1.25 to 1.51), indicating strong practical relevance.

The longitudinal trends for the Simple Smile Index, Weighted Smile Index, and dental exposure are shown in Figures 1–3. Group comparisons are summarized in Figures 4–6, while distributional differences are presented in Figures 7–9. Table 1 summarizes the principal numerical outcomes reported in the manuscript.

Correlation analysis revealed strong positive associations between year and both composite indices (rho = 0.69) and dental exposure (rho = 0.69), confirming a steady temporal trend. AU12 activation also showed a moderate but significant increase (rho = 0.48). No statistically significant differences were observed in Duchenne versus non-Duchenne smile distribution, suggesting that changes primarily involved intensity rather than qualitative type [21–22].

Discussion

The results of this study indicate a significant evolution in smile characteristics over the past four decades, with a clear trend toward increased expressivity and dental display. From a stomatological perspective, these findings align with established principles emphasizing the importance of incisal display, smile arc, and lip dynamics in smile attractiveness [1–2,9,11,15,18,20]. The increase in dental exposure may reflect a shift toward aesthetic ideals that favor visibility of the dentition, often associated with youthfulness and vitality [1–2,9,18]. Additionally, greater activation of AU12 suggests more pronounced engagement of the zygomaticus major muscle, contributing to broader and more expressive smiles [21–22].

The influence of social media represents a plausible explanatory factor, although the present study cannot demonstrate causality [24–25]. Digital platforms reward facial expressions that are immediately recognizable and visually engaging, potentially promoting more visible smile configurations. This phenomenon may be particularly relevant in contexts such as beauty pageants, where visual performance is central.

The integration of AI-assisted FACS analysis constitutes an important methodological advancement. Traditional FACS coding is labor-intensive and subject to inter-rater variability, whereas AI systems offer scalable and standardized alternatives [21,23]. This approach may have future applications in orthodontic diagnostics, smile design, and aesthetic evaluation; however, the present findings should be interpreted as image-based measurements rather than direct clinical assessments [11,23,26].

Limitations

The study is limited by its relatively small sample size and reliance on photographic material, which may vary in quality and standardization. In particular, the retrospective nature of the dataset implies substantial heterogeneity in image acquisition across decades, including differences in resolution, lighting conditions, camera optics, and image compression. Earlier images from the 1980s and 1990s are more likely to present lower resolution and contrast, potentially affecting the accurate identification of facial action units and dental exposure.

A further limitation concerns the lack of a standardized head-position and smile-elicitation protocol. The images analyzed were not obtained under controlled clinical conditions; therefore, variations in head posture, camera angle, lip posture at rest, and the degree of posed versus spontaneous smiling may have influenced the measurements. In orthodontic analysis, even minor deviations in head orientation or lip dynamics can significantly alter the perceived incisor display and smile arc [11]. Consequently, part of the observed variability may reflect differences in photographic conditions rather than true anatomical or functional variation.

Additionally, the study population, comprising exclusively Miss Italia winners, represents a highly selected, non-representative sample characterized by specific aesthetic, anthropometric, and socio-cultural criteria. As such, the findings cannot be generalized to the broader population; rather, they reflect trends within a highly curated group subject to visual and media-driven selection pressures. The use of a 2004 temporal cut-off should also be interpreted cautiously. This division represents a conceptual proxy for the emergence of social media rather than a discrete biological or behavioral threshold [24–25].

Finally, although artificial intelligence-based FACS scoring enhances internal consistency and reproducibility, it does not replace clinical evaluation. The quality and variability of input data inherently limit AI systems. They may not fully capture three-dimensional dynamics, muscle tone, or functional aspects of the smile that are routinely assessed in clinical orthodontics [11,21,23]. Static image analysis cannot replicate dynamic smile evaluation, which remains essential for comprehensive diagnosis [11].

Conclusions

The results of this study on Miss Italia winners between 1982 and 2025 suggest a clear, statistically significant transformation in the expressivity of posed smiles. The period identified as the social media era (2004–2025) was characterized by higher smile intensity compared with the preceding two decades. Specifically, the increase in the Simple Smile Index and the Weighted Smile Index, from 15.91 to 19.59, suggests a trend toward more emphasized visual communication.

This phenomenon was primarily driven by the dental exposure component, which showed the most marked change and had the largest effect size. These data indicate an aesthetic shift toward wider, more dynamic posed smiles, which may be better suited to photographic rendering and the rapid visual consumption typical of digital devices and social platforms. Despite the increase in overall intensity, the proportion of Duchenne smiles showed no significant changes between the two eras (p = 0.469), suggesting that while baseline expressivity remained high, the visual performance of the smile may have evolved in response to new photogenic standards. These conclusions should be interpreted with caution because the retrospective design does not allow causal inference regarding social media exposure, orthodontic treatment, or other sociocultural factors.

Declarations