PREDICAL – AN INNOVATIVE AI TOOL FOR PREDICTING POSTOPERATIVE CALCIUM REQUIREMENT IN PRIMARY HYPERPARATHYROIDISM Dinesh.goli@gmail.com

 

PREDICAL – AN INNOVATIVE AI TOOL FOR PREDICTING POSTOPERATIVE CALCIUM REQUIREMENT IN PRIMARY HYPERPARATHYROIDISM

3

India

Oral only

Postoperative hypocalcemia is a common and potentially serious complication following parathyroidectomy. Traditional monitoring delays discharge decisions, making same-day surgery challenging. With the advent of artificial intelligence (AI), predictive models can help stratify risk early and guide discharge planning.

AIM-To develop and validate an AI-based machine learning model that predicts calcium requirement and supports safe same-day discharge in primary hyperparathyroidism. A retrospective cohort of 840 patients who underwent parathyroidectomy for primary hyperparathyroidism was analyzed. Variables included preoperative calcium, PTH, vitamin D, ALP, bone mineral density and postoperative PTH. Data were split into training and testing sets (80:20). An AI-based logistic regression model using python in Google Colab platform was trained to predict the requirement of postoperative calcium supplementation. Model performance was assessed using accuracy, sensitivity, and specificity

The AI-based model demonstrated a high overall accuracy of 90%, with a sensitivity of 80% and specificity of 92% in predicting the need for postoperative calcium supplementation. It effectively identified patients at both high and low risk for hypocalcemia, making it a reliable tool for perioperative risk stratification. The model’s ability to accurately rule out low-risk individuals supports its use in facilitating early discharge and assessing suitability for day care parathyroidectomy

The PrediCAL AI model demonstrates strong potential in identifying low-risk patients for hypocalcemia following parathyroidectomy. Its high specificity supports its application in day care surgery planning for primary hyperparathyroidism. This model will form a base to develop better Machine learning models for application in Endocrine Surgery