Calcium ion dynamic trajectory is associated with prognosis in patients with sepsis: A potential class mixture...

Disturbed calcium homeostasis in patients with sepsis is associated with poor prognosis; however, its dynamic pattern and clinical significance remain unclear. This study aimed to establish an early warning system for calcium homeostasis in sepsis by analyzing the association between calcium ion dynamics and clinical outcomes. This retrospective cohort study enrolled 949 patients with sepsis from June 2018 to February 2025, to analyze the longitudinal trajectory of serum calcium levels within 5 days after admission based on the latent class mixture model (LCMM). The goodness of fit of the group was assessed using the Akaike Information Criterion, Bayesian Information Criterion, and entropy value. An XGBoost machine learning model was constructed to identify high-risk patients. Three trajectory subclasses were identified: stable normal (Class 1), persistently low calcium (Class 2), and rapid decline (Class 3). Multifactorial Cox regression analysis showed that calcium ion dynamic trajectory was an independent predictor of in-hospital mortality, with the highest risk of death in the rapidly declining group, followed by the persistently low calcium group. The calcium concentration showed a U-shaped relationship with the risk of death, with 1.75-2.25 mmol/L as the safe interval, whereas either low or high calcium concentrations significantly increased the risk. The classifier constructed based on XGBoost was optimized using 5-fold cross-validation and effectively identified high-risk patients. This study revealed, for the first time, the dynamic characteristics of calcium metabolism in sepsis based on the latent class mixture model, providing a novel basis for prognostic stratification and precise intervention.