Development and validation of a predictive model for complications in robotic thyroid surgery

Robotic thyroid surgery (RTS) has gained popularity due to its minimal invasiveness and improved cosmetic outcomes. However, complications like hypoparathyroidism, bleeding, and infection can occur before surgeons' learning curve completion. The purpose of present research is to develop and validate a predictive model for complications in RTS prior to their learning curve completion. We retrospectively analyzed data from 236 cases accepted RTS at our institution from Jan 2020 to Dec 2022. The data of included cases were classified into training set (n = 165) and validation set (n = 71). Data on clinical characteristics, surgical details, and postoperative outcomes were collected. Several regression analysis models were applied to identify independent factors of complications. A nomogram was constructed and validated using the training and validation sets. The overall complication rate was 29.24% (69/236 patients). Multivariate analysis identified BMI (odds ratio [OR] = 3.52, 95% Confidence Interval [CI]: 1.179-11.05), Hashimoto's thyroiditis (OR = 1.69, 95% CI: 1.44-2.05) and tumor size (OR = 2.37, 95% CI: 1.10-5.18) as independent predictors of complications. Based on these predictors, nomogram was constructed. The C-index was 0.8574 (95% CI: 0.762-0.9529) in validation set, indicating good predictive accuracy. This study proposed and validated a nomogram model to predict complications in RTS prior to their learning curve completion. This model can assist clinicians in preoperative counseling and decision-making to minimize surgical risks.