Development and validation of an AI model predicting post-extraction complications: A multicenter cohort study protocol

Abstract

Dry socket (alveolar osteitis) and post-operative surgical-site infection (SSI) are most common complications affecting 1-20% extraction sites. Risk factors include smoking, diabetes, prolonged surgery and operative trauma. A multicenter cohort using retrospective derivation (n≈2000, 3–5 centers) and prospective validation (12–18 months, ≥2 centers) will be done. The patients included will be adults ≥18 years undergoing extraction and outcome assessed will be dry socket (pain, exposed bone days 1–5); SSI (purulence/swelling within 30 days). The expected outcome model should be able to predict the prognosis of the dry socket and SSI, thus reducing the injudicious use of antibiotics and increase in expertise of the surgeon.