The Ongoing Evolution of AI in Craniofacial Surgery: From Theory to Reality and Beyond

Artificial intelligence (AI) has had a long-standing appeal in craniofacial surgery. To examine its ongoing evolution in the field, this article outlines theories, current applications, and future directions. Current AI, machine learning (ML), and deep learning (DL) models associated with craniofacial surgery are presented and categorized based on the perioperative stage they can assist with. These applications include prenatal diagnosis of cleft lip and palate, cephalometric analysis and diagnosis of craniosynostosis, segmenting CT scans and detecting fractures, predictive analytics, surgical planning, automated computer-aided design (CAD) for patient-specific guides and custom implants, augmented reality (AR)-assisted intraoperative navigation, as well as automated postoperative outcome assessment and patient education. Most of these advances remain to be broadly integrated in routine surgical practice. Notable barriers to integration include limited datasets, concerns about technical accuracy and margins of clinical acceptability, as well as insufficient external validation or generalizability, transparency, and ethics. Potential use cases for AI to improve craniofacial care are discussed. Taking the successes and barriers into account, future progress and uptake of these methods will depend on federated learning, generative design, and explainable AI (XAI).