Currently accepted at: JMIR Aging
Date Submitted: Jun 3, 2025
Open Peer Review Period: Jun 10, 2025 - Aug 5, 2025
Date Accepted: Feb 27, 2026
(closed for review but you can still tweet)
This paper has been accepted and is currently in production.
It will appear shortly on 10.2196/78495
The final accepted version (not copyedited yet) is in this tab.
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Integrated High-Throughput Targeted Metabolomics and Machine Learning for Early Prediction and Prevention of Postoperative Delirium in Elderly Surgical Patients
ABSTRACT
Background:
Postoperative delirium (POD) in elderly hip fracture patients is associated with high morbidity and severe adverse outcomes, yet its pathogenesis remains unclear.
Objective:
This study aimed to develop a predictive model for POD following hemiarthroplasty in elderly patients by integrating high-throughput targeted metabolomics and machine learning, enabling early identification and intervention for high-risk individuals to improve postoperative recovery.
Methods:
In this prospective, observational, multi-center cohort study, 245 elderly patients undergoing hemiarthroplasty for hip fractures were enrolled. Perioperative cognitive assessments and clinical data were collected, with preoperative blood samples analyzed via high-throughput targeted metabolomics. Machine learning algorithms were employed to identify metabolomics signatures associated with POD. Differential metabolites were screened using Random Forest (RF) and Lasso regression (Least Absolute Shrinkage and Selection Operator). Predictive models were constructed using Gradient Boosting, Logistic Regression, and Random Forest. Model performance was evaluated by Receiver Operating Characteristic (ROC) curves and area under the curve (AUC).
Results:
Absolute quantification of 201 metabolites revealed 41 significantly differentially expressed metabolites between POD and non-POD groups (P < 0.05). RF and Lasso regression identified 16 candidate biomarkers for model construction. The Logistic Regression model demonstrated optimal performance, achieving an AUC of 0.855 (95% CI: 0.8–0.91) in the overall cohort. Upon 7:3 random partitioning into training and test sets, the model maintained robust predictive accuracy with AUCs of 0.844 and 0.856.
Conclusions:
Integration of preoperative metabolomics profiling and machine learning enables accurate preoperative or early postoperative prediction of POD in elderly hip fracture patients. This approach facilitates personalized risk stratification and tailored clinical management, potentially reducing complications and enhancing recovery outcomes. The model highlights the translational potential of metabolomics biomarkers combined with artificial intelligence for precision medicine in geriatric perioperative care. Clinical Trial: Chinese Clinical Trial Registry ChiCTR-CPC-15006141; https://www.chictr.org.cn/ indexEN.html
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.