Artificial Intelligence-Augmented Morphometric Reporting for Standardized Facial Aesthetic Treatment Planning: A Multi-Rater Reliability Study.

BACKGROUND

Facial aesthetic assessment is inherently subjective, leading to significant inter-observer variability in treatment planning. Artificial intelligence (AI) offers a potential framework for standardized morphometric reporting.

OBJECTIVE

To evaluate whether AI-augmented morphometric reports improve the consistency of simulated aesthetic treatment plans among dermatologists.

METHODS

Six dermatologists (3 experts, 3 residents) proposed treatment plans (neurotoxin units and filler volumes) for 76 standardized frontal images from the Chicago Face Database. In Phase I (Baseline), raters used visual intuition. In Phase II (AI-Augmented), raters were provided with objective morphometric reports. The primary outcome was the change in inter-rater reliability, measured by the Intraclass Correlation Coefficient (ICC(2,1)).

RESULTS

Significant baseline asymmetry (> 1.5 scaled units) was present in 58.4% of cases. The overall ICC(2,1) improved from 0.62 (95% CI: 0.54-0.69) to 0.88 (95% CI: 0.81-0.93) following AI augmentation. The most substantial improvement was observed in the resident group (ICC: 0.46 to 0.84). The mean absolute deviation from the expert-group mean declined from 0.34 to 0.12 scaled units.

CONCLUSION

AI-augmented morphometric reporting was associated with improved inter-rater agreement and reduced variability in simulated facial aesthetic treatment planning, particularly among less experienced raters. However, improved agreement does not necessarily reflect improved clinical accuracy, and future validation against clinical outcomes is warranted. The AI system functioned as a decision-support tool rather than a replacement for clinician judgment.