Biological Variability and Biomedical Innovation in Personalized Orthodontics: A Systematic Review with Narrative Synthesis

Abstract

Background/Objectives: Orthodontic treatment response varies considerably among patients because mechanical loading interacts with individual biological, anatomical, systemic, and behavioral characteristics. Advances in mechanobiology, molecular biomarkers, omics technologies, and artificial intelligence (AI) may support a more personalized approach to treatment. This systematic review (SR) aimed to evaluate human clinical evidence concerning these domains and their potential contribution to predicting orthodontic response, improving clinical decision-making, and reducing adverse effects. Methods: A systematic search of PubMed, Scopus, and Web of Science was conducted through June 2026. Primary studies involving human participants and addressing orthodontic tooth movement (OTM), biological variability, biomarkers, genomics, epigenetics, multi-omics, or AI-supported orthodontics were selected according to predefined eligibility criteria. The review was conducted and reported in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020, and heterogeneous evidence was synthesized narratively across four predefined domains. The review protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration number CRD420261486348). Risk of bias was assessed using design-appropriate instruments. Certainty of evidence was assessed at the outcome level using GRADE principles adapted to narrative synthesis. Results: The search identified 1051 records, of which 14 primary human studies were included. The clinical evidence comprised randomized trials, longitudinal and cross-sectional biomarker studies, genetic association studies, and a case report. Molecular and multi-omics studies identified candidate markers of biological response and patient phenotype. The included computational studies examined genetic association rules, gingival-enlargement classification, and AI-assisted root-resorption measurement; none demonstrated AI-guided selection of orthodontic forces. However, small samples, methodological heterogeneity, limited adjustment for confounding, and lack of external validation restricted clinical translation. Certainty of evidence supporting clinical personalization applications was predominantly very low. Conclusions: Personalized orthodontics should remain grounded in comprehensive diagnosis, biologically compatible mechanics, and individualized monitoring. Molecular and computational tools may complement clinical assessment, but routine implementation requires standardized protocols, multicenter validation, and prospective evidence of patient-relevant benefit.