无题作品

This study developed and validated a risk prediction model and a personalized advantage index (PAI) model for postoperative sleep disturbance (PSD) in cardiac surgery patients, and established a model-driven personalized intervention protocol. A prospective study (Nov 2023–May 2024) used logistic regression to identify PSD risk factors. Separate models for non-thoracotomy and thoracotomy patients were evaluated by AUC, sensitivity, and specificity. The PAI model was built using random forest and L1-regularized regression, assessed with multiple metrics. A subsequent stratified RCT (70 patients, Nov 2024–Jan 2025) compared intervention and control groups, with PSD incidence, PSQI scores, and adverse reactions as outcomes.