Publication Details
Abstract
Background. Hungry Bone Syndrome (HBS) remains one of the most clinically significant forms of severe postoperative hypocalcemia after parathyroidectomy. The absence of standardized preoperative risk stratification limits early prevention and individualized postoperative monitoring. Objective. To develop and perform pilot internal validation of a digital clinico-biochemical model for predicting HBS based on preoperative variables. Materials and methods. A training cohort of 200 observations was analyzed; HBS was registered in 20 patients (10.0%). Predictors included the type of hyperparathyroidism, intact parathyroid hormone level, alkaline phosphatase activity expressed as a multiple of the upper limit of normal, corrected calcium, and an integrated digital score. Discriminative ability was assessed by ROC analysis; time-dependent prognostic significance was evaluated using Kaplan-Meier analysis and Cox regression. Results. The median score was higher in patients with HBS than in those without HBS: 14.75 [13.70-15.62] versus 8.10 [5.88-10.50] points (p<0.001). The area under the ROC curve was 0.961 (bootstrap 95% CI 0.927-0.986). The operational high-risk group included 30 patients; HBS developed in 17 of them (56.7%) versus 3 of 170 (1.8%) outside the high-risk group. Sensitivity of the high-risk flag was 85.0%, specificity was 92.8%, and negative predictive value was 98.2%. Kaplan-Meier analysis demonstrated significant separation between risk groups (log-rank chi-square=100.284; p<0.001). In the Cox model, each additional score point was associated with a 59.4% increase in event hazard (HR=1.594; p=0.001). Conclusion. The proposed digital model demonstrated strong internal discrimination and a marked ability to identify patients with clinically meaningful HBS risk. These results should be considered pilot findings and require external prospective validation.