TITLE:
AI-Assisted Case Study of Delirium in ICU Patients: Predictive Analysis, Monitoring, and Interventions
AUTHORS:
Rocco de Filippis, Abdullah Al Foysal
KEYWORDS:
Delirium Prediction, ICU Patient Monitoring, Machine Learning in Healthcare, Random Forest Model, Real-Time Risk Assessment
JOURNAL NAME:
Open Access Library Journal,
Vol.12 No.1,
January
23,
2025
ABSTRACT: Delirium is a critical condition affecting a significant proportion of ICU patients, often resulting in prolonged hospital stays, cognitive decline, and increased mortality. Early identification and management of delirium are essential to improve patient outcomes, but traditional methods are often reactive and resource intensive. This study leverages machine learning, specifically a Random Forest model, to predict delirium risk using simulated ICU patient data. Key physiological features, including sedation levels, cardiovascular stress, and oxygen saturation, were identified as the most influential predictors. Although the model achieved moderate accuracy (50%), it provided meaningful insights into risk patterns and demonstrated the utility of real-time monitoring in guiding timely and targeted interventions. Dynamic risk fluctuations, visualized through patient simulations, highlighted the importance of continuous monitoring over static assessments. Figures and tables illustrate model performance, feature importance, and real-time monitoring trends, offering actionable insights for clinical applications. This paper underscores both the promise and the challenges of integrating AI-assisted tools into ICU workflows, paving the way for future research to refine predictive accuracy and enhance practical deployment in clinical settings.