Analysis of Preschool Mental Health based on FCM-PSO Clustering Algorithm

Mental wellbeing assessment remains challenging in early childhood because of the evolving nature of behavioural indicators in preschoolers. Standard screening methods often fail to capture patterns. The key problem lies in effectively analysing complex, multidimensional behavioural data where symptoms frequently overlap across different developmental conditions. This study proposes an enhanced Fuzzy C-Means (FCM) clustering approach optimized with Particle Swarm Optimization (PSO) to address these challenges. The algorithm processes teacher assessments, parent reports, and observational data to identify distinct mental health profiles while accommodating the inherent ambiguity in childhood behaviours. Evaluation metrics demonstrate exceptional performance: silhouette score (0.96) confirms near-perfect cluster cohesion, Davies-Bouldin index (0.18) indicates outstanding separation, and classification accuracy reaches 96.7% when validated against clinical assessments. The system achieves 95.8% precision in identifying anxiety markers and 96.2% recall for detecting social withdrawal patterns. These results significantly outperform traditional K-means clustering (accuracy=81.3%) and hierarchical methods. By providing quantifiable, interpretable clusters of mental health status, this FCM-based approach enables early, accurate identification of at-risk children while reducing subjective bias in preschool mental health screening. The framework's adaptability to various data sources and exceptional performance metrics (F1-score=96.0%) demonstrate clinical-grade reliability for implementation in educational and healthcare settings.