Integrated High-Throughput Targeted Metabolomics and Machine Learning for Early Prediction and Prevention of Postoperative Delirium in Older Adult Surgical Patients: Prospective Multicenter Cohort Study.

BACKGROUND

Postoperative delirium (POD) is a common and severe complication in older adult patients with hip fracture, yet its pathogenesis remains unclear.

OBJECTIVE

This study aimed to develop a predictive model for POD following hemiarthroplasty in older adult patients by integrating high-throughput targeted metabolomics and machine learning.

METHODS

In this prospective multicenter cohort study, 260 older adult patients undergoing hemiarthroplasty for hip fracture were enrolled. Preoperative serum samples were analyzed via high-throughput targeted metabolomics. Differential metabolites were screened using random forest (RF) and least absolute shrinkage and selection operator regression. Predictive models were constructed using gradient boosting, logistic regression, and RF, with performance evaluated using receiver operating characteristic curves and the area under these curves (area under the receiver operating characteristic curve [AUC]).

RESULTS

Absolute quantification of 201 metabolites revealed 41 (20.4%) significantly differentially expressed metabolites. RF and least absolute shrinkage and selection operator regression identified 16 candidate biomarkers. The logistic regression model demonstrated optimal performance, achieving an AUC of 0.855 (95% CI 0.800-0.910) in the overall cohort. Upon 7:3 partitioning into training and test sets, the model maintained robust predictive accuracy, with AUCs of 0.844 and 0.856, respectively.

CONCLUSIONS

Integration of preoperative metabolomics and machine learning enabled accurate prediction of POD in older adult patients with hip fracture, facilitating personalized risk stratification and tailored clinical management.