Predicting postoperative cognitive dysfunction in older cardiac surgery patients: An integrated machine learning approach with a visual nomogram

Objectives This study integrated machine learning algorithms to identify key risk factors for postoperative cognitive dysfunction (POCD) in older cardiac surgery patients. This study aimed to develop a predictive nomogram to assist clinicians and nurses in identifying high-risk patients and implementing targeted interventions. Methods A prospective cohort study was conducted with 353 older cardiac surgery patients admitted to the surgical intensive care unit (ICU). Data on demographics, laboratory results, and clinical characteristics were collected. The least absolute shrinkage and selection operator (LASSO) regression was applied to determine the most relevant predictors for POCD. These predictors were incorporated into a multivariate logistic regression model to construct a predictive nomogram. Model performance was assessed using receiver operating characteristic (ROC) curve analysis, calibration curves, and decision curve analysis. Results POCD was observed in 49.86% of patients. Seven independent predictors were identified: surgical approach, pre-existing comorbidities, operation duration, intraoperative blood loss, sleep quality score during ICU stay, Acute Physiology and Chronic Health Evaluation II (APACHE II), and self-care ability. These predictors were incorporated into the predictive nomogram; it demonstrated robust predictive performance with an area under the ROC curve (AUC) of 0.786. The nomogram exhibited excellent calibration and discrimination. Decision curve analysis confirmed its clinical utility across a broad range of threshold probabilities. Conclusions A precise and effective nomogram was developed using the surgical approach, Underlying comorbidities, operation duration, blood loss, ICU sleep quality, APACHE II, and self-care ability as predictors of POCD in older cardiac surgery patients. Implications for Clinical Practice This nomogram provides a valuable tool for early detection and prevention of POCD, enabling clinicians to make informed decisions and tailor interventions. Its application can help reduce the incidence of POCD, ultimately improving patient outcomes and quality of care.