AI-Supported, Integrative Prediction of Postoperative Delirium: Protocol for the CONFUSED Study

Area of Science:

• Geriatric Medicine
• Surgical Oncology
• Neuroscience

Background:

• Postoperative delirium (POD) is a significant complication in elderly surgical patients, leading to adverse outcomes like extended hospitalization and increased mortality.
• The precise pathophysiological mechanisms of POD remain unclear, and effective biomarkers for early detection and prevention are currently unavailable.

Purpose of the Study:

• To identify molecular and clinical predictors of POD by integrating multimodal data.
• To develop predictive models for POD using proteomic, transcriptomic, and epigenetic analyses combined with clinical information.
• To identify delirium-associated genes, proteins, epigenetic signatures, and at-risk patient subgroups.

Main Methods:

• A prospective observational cohort study involving adult patients undergoing major surgery under general anesthesia.
• Collection of blood samples at four time points for proteomic, transcriptomic, and epigenetic analyses.
• Systematic recording of clinical data and delirium assessments using the Confusion Assessment Method (CAM).
• Application of univariate, multivariate, and machine learning statistical methods for biomarker identification and model development.

Main Results:

• The study is ongoing, with recruitment initiated in January 2024 and 135 patients enrolled to date.
• Sample collection and laboratory analyses are in progress, with data analysis commencing in January 2026.
• First results are anticipated in July 2026, following the completion of recruitment and data analysis.

Conclusions:

• The CONFUSED study seeks to enhance POD prediction and understanding through multimodal data integration and machine learning.
• Expected outcomes include the development of personalized preventive strategies for POD.
• The findings are anticipated to contribute to improved perioperative care for at-risk surgical patients.