Resumo

Objectives: To evaluate the feasibility and clinical utility of an AI-based system for analyzing dynamic facial expressions and generating personalized injection maps for incobotulinum toxin. Specifically, we aimed to determine whether AI-guided planning could optimize dosing, improve facial symmetry, and reduce retreatment needs compared to traditional standardized injection protocols, while maintaining safety and high patient satisfaction. This project seeks to establish a framework for integrating AI into clinical practice for improvement and precision of toxin procedures. Introduction: Botulinum toxin treatments are often based on standardized injection points, overlooking individual facial dynamics. This can lead to asymmetry, suboptimal dosing, or overcorrection. Incobotulinum toxin, free of accessory proteins, is well suited for repeated and precise protocols. Artificial intelligence (AI) applied to high-resolution video analysis of facial expressions offers the potential to identify muscle hyperactivity, asymmetry, and contraction intensity, enabling a new era of truly personalized botulinum toxin planning. Materials / method: A prospective pilot study enrolled 20 patients seeking upper-face botulinum toxin treatment. Standardized high resolution videos were recorded including neutral, frown, smile, eyebrow elevation, and eye closure. An AI model combining facial landmark tracking and facial action unit analysis generated personalized injection maps indicating sites and suggested doses. Patients were randomized to AI-based or conventional standardized mapping with incobotulinum toxin. Outcomes included symmetry indices, retreatment rates, GAIS, and patient satisfaction. Results: 20 patients were included (10 AI-guided, 10 conventional). Follow up at 2 weeks and 12 weeks, the AI group showed a 28% reduction in retreatment compared with conventional. Symmetry indices improved significantly in the AI group, with fewer visible asymmetries on dynamic expression. Patient satisfaction scores were higher (Likert 4.7 vs 3.9). GAIS evaluations confirmed superior natural appearance in the AI-guided cohort. No adverse events were observed in either group. Conclusion: This pilot study supports AI-guided injection mapping as a promising tool to enhance precision, symmetry, and patient satisfaction in incobotulinum toxin treatments. While results are encouraging, this remains an exploratory model. Larger and more diverse populations are needed to validate findings and establish clinical guidelines. AI-assisted planning may become an effective tool to improve daily consultations, optimize dosing, and increase efficiency in aesthetic practice.