Advances in predictive modeling of cranial growth in craniosynostosis: A systematic review and meta-analysis.

BACKGROUND: Craniosynostosis, characterized by premature fusion of cranial sutures, disrupts normal skull growth and may lead to neurological and functional complications. Despite advances in surgery, planning remains largely experience-based due to limited patient-specific predictive tools. METHODS: A PRISMA-compliant systematic review was conducted across PubMed/MEDLINE, Scopus, and Web of Science (January 2020 to November 2025). Eligible studies enrolled pediatric craniosynostosis patients using predictive or morphometric modeling. Risk of bias was assessed using PROBAST, QUIPS, and CASP. A random-effects meta-analysis was performed where applicable. RESULTS: Twenty-nine studies were included, spanning FEA (n = 14), statistical shape and regression models (n = 9), and machine learning approaches (n = 3). Meta-analysis of four studies reporting cephalic index showed a non-significant pooled effect (MD = -1.23; 95% CI -3.35 to 0.89; p = 0.26; I2 = 57%), with moderate heterogeneity limiting conclusions. Meta-analysis of two studies reporting intracranial volume showed a significant reduction (SMD = -1.81; 95% CI -3.32 to -0.31; p = 0.02; I2 = 77%), though the high heterogeneity warrants cautious interpretation given the very limited number of contributing studies. Most studies (69%) were rated as moderate risk of bias, with common limitations including small single-center samples, retrospective designs, and limited external validation. CONCLUSIONS: FEA, statistical shape models, and ML approaches each offer complementary strengths for surgical planning in craniosynostosis. Future research should prioritize multicenter validation, radiation-free imaging, and standardized reporting to enable patient-specific clinical deployment.