PREVALENCE AND RISK FACTORS OF POSTOPERATIVE INFECTIONS IN CARDIAC SURGERY PATIENTS USING AI ASSESSMENT

Abstract

Background: Surgical site infections (SSIs) are among the most common and costly complications following cardiac surgery, significantly affecting morbidity and mortality rates. Traditional risk assessment models often fail to capture complex interactions among variables, limiting predictive precision. Recent advances in artificial intelligence (AI) offer opportunities to enhance infection prediction and prevention. Objective: To assess the prevalence of surgical site infections in cardiac surgery patients and identify associated risk factors using AI-driven analysis tools. Methods: This cross-sectional study was conducted over eight months (June 2024–February 2025) at two tertiary care cardiac centers. A total of 600 adult patients undergoing cardiac surgery were enrolled based on defined inclusion and exclusion criteria. Data were collected from electronic health records and included demographic, preoperative, intraoperative, and postoperative variables. Surgical site infections were diagnosed using CDC criteria and validated through independent clinical review. Statistical analysis included univariate and multivariate logistic regression, while machine learning models—Random Forest and Gradient Boosting—were developed to assess predictive accuracy. Results: SSIs occurred in 14% of patients (n=84), with superficial incisional infections being most common. Diabetes (OR 2.5), obesity (OR 2.1), surgery duration >5 hours (OR 3.2), re-exploration (OR 4.0), and prolonged ventilation (OR 3.5) were significant independent predictors. Gradient Boosting demonstrated superior predictive performance with an AUC-ROC of 0.91 compared to 0.89 for Random Forest. Conclusion: The integration of AI models enhances the predictive accuracy of SSI risk stratification in cardiac surgery. Early identification of high-risk patients through AI tools can support targeted prevention strategies and improve surgical outcomes.

Ähnliche Arbeiten