Predicting Postoperative Delirium in Older Patients Before Elective Surgery: Multicenter Retrospective Cohort Study

Area of Science:

• Geriatric Medicine
• Artificial Intelligence in Healthcare
• Surgical Risk Assessment

Background:

• Increasing rates of elective surgeries in older adults necessitate improved risk stratification.
• Postoperative delirium (POD) is a significant concern in this demographic, impacting patient outcomes.
• Machine learning (ML) offers potential for enhanced predictive modeling in surgical patient management.

Purpose of the Study:

• To evaluate the predictive performance of ML models for POD in older surgical patients.
• To determine the added value of preoperative neuropsychological assessments in predicting POD.

Main Methods:

• Retrospective cohort study of 1624 older adults (≥70 years) undergoing elective surgery.
• Data included sociodemographic, clinical, surgical, and neuropsychological variables.
• ML models were assessed using AUC, with Shapley Additive Explanations for feature importance.

Main Results:

• A random forest model achieved an AUC of 0.760 for POD prediction.
• Including all pre- and intraoperative features slightly improved AUC to 0.783 (P=.24).
• Specific neuropsychological tests (Montreal Cognitive Assessment, Trail Making Tests) were key predictors, enhancing model accuracy.

Conclusions:

• Preoperative ML-based risk prediction for POD can improve surgical planning and patient care.
• Integrating cognitive assessments into ML models enhances the prediction of POD in elderly surgical patients.
• Early identification of high-risk individuals facilitates targeted interventions to mitigate delirium.