SMARTPHONE-DERIVED FACIAL ANTHROPOMETRIC MEASUREMENTS FOR PREDICTION OF DIFFICULT AIRWAY: A COMPARISON WITH CONVENTIONAL AIRWAY ASSESSMENT METHODS

Background: Preoperative recognition of difficult airway remains central to safe anesthetic care because failed or delayed airway control may rapidly compromise oxygenation. Conventional bedside tests are simple and widely used, but their accuracy is often limited by subjective interpretation and observer variability. Smartphone-based facial imaging may offer a practical way to capture objective craniofacial measurements related to laryngoscopic view, especially in settings where advanced airway-assessment technology is not routinely available for daily preoperative screening practice locally. Objective: To evaluate smartphone-derived facial anthropometric measurements for predicting difficult laryngoscopy and compare their diagnostic performance with conventional airway assessment methods. Methods: This prospective observational comparative study included 35 adults aged 18–65 years, ASA I–III, undergoing elective surgery under general anesthesia with endotracheal intubation. Conventional assessment included Mallampati class, mouth opening, thyromental distance, sternomental distance, neck circumference, upper lip bite test, and neck extension. Standardized smartphone photographs were analyzed for facial width, mandibular width, chin projection angle, neck-chin angle, neck-to-face ratio, jawline angle, and submandibular contour. Difficult laryngoscopy was defined as Cormack–Lehane Grade III–IV. Analysis included logistic regression, ROC curves, diagnostic indices, leave-one-out cross-validation, bootstrap validation, and calibration assessment. Results: Difficult laryngoscopy occurred in 7 patients (20.0%), while 28 patients (80.0%) had easy laryngoscopy. Chin projection angle showed apparent AUC 0.90 and LOOCV AUC 0.85. Neck-chin angle showed apparent AUC 0.86 and LOOCV AUC 0.81. Thyromental distance had AUC 0.83, sensitivity 85.7%, and specificity 75.0%. The combined model achieved apparent AUC 0.94, LOOCV AUC 0.89, optimism-corrected AUC 0.87, sensitivity 85.7%, specificity 90.0%, accuracy 88.6%, calibration slope 0.72, and Brier score 0.12. Conclusion: Smartphone-derived facial anthropometric measurements appeared useful as adjuncts to conventional airway assessment. Their clinical use requires larger multicenter validation before routine implementation. Keywords: Airway Management; Anesthesiology; Anthropometry; Intubation, Intratracheal; Laryngoscopy; Preoperative Care; Smartphone.