Efficacy of an Intelligent Computational Nursing Decision Support System in Precision Critical Care for Delirium Prevention: A Randomized Controlled Trial

Delirium affects up to 83% of ventilated ICU patients, necessitating precise non-pharmacological interventions like the SCAN-D program, which aims to improve outcomes by integrating smart computing and virtual reality into nursing decisions. This three-year, assessor-blinded, parallel randomized controlled trial will enroll 188 patients, assigning them to either usual care or the SCAN-D group, where the latter receives personalized strategies based on machine learning risk weights, including family-supervised immersive VR exercise and cognitive training. Throughout their stay, patients will be monitored via shift-based assessments (ICDSC/RASS) and nightly sleep tracking using actigraphy combined with one-lead EEG, while long-term cognitive function and quality of life (EQ-5D-3L) will be evaluated at three and six months post-discharge. The primary outcomes are the incidence of delirium and delirium-free days within the first 28 days of ICU admission, measured using the ICU Delirium Screening Checklist. Secondary outcomes include sleep quality and quantity, functional status during the ICU stay, ICU length of stay, duration of mechanical ventilation, total hospital length of stay, ICU mortality rate, Sedative and Analgesic Loading, 90-day Mortality Post-discharge, cognitive function, and quality of life. Finally, the collected data will be analyzed using SPSS 22.0 through independent t-tests, Chi-square tests, and Generalized Estimating Equations (GEE) to determine the program's efficacy in reducing delirium incidence and enhancing overall patient recovery.