Eye Movements to Predict Post-Operative Delirium (EYE-POD)

This prospective trial evaluates the integration of automated eye movement assessments as a novel, noninvasive risk biomarker for postoperative delirium (POD) in older adults. Upon enrollment, trained clinical staff will collect baseline demographic data and clinical characteristics. Chronic medical burden will be quantified using the Charlson Comorbidity Index (CCI). To establish baseline cognitive function and control for preexisting cognitive impairment, all participants will undergo the Montreal Cognitive Assessment (MoCA).

Technical Protocol for Eye Movement Recording

Preoperatively, participants will complete an automated eye movement tracking protocol consisting of two distinct paradigms:

Prosaccadic Task: Utilized to measure specific ocular motor parameters, including saccadic latency, velocity, and accuracy.

Free Visual Exploration Task: Utilized to capture free-gaze metrics and spatial distribution patterns.

All eye-tracking data will be recorded using automated software to ensure objectivity and reproducibility.

Postoperative Monitoring Following the surgical procedure, patients will be systematically monitored for the development of POD. The Confusion Assessment Method (CAM) will be administered by trained examiners twice daily (morning and evening) for the first three postoperative days (72 hours) to detect and document the incidence and severity of delirium.

Statistical and Predictive Objectives

The collected data will be analyzed to fulfill three primary scientific objectives:

1. Parameter Association: To identify which specific preoperative eye-tracking metrics (latency, velocity, accuracy, or free-gaze variables) demonstrate a statistically significant association with the incidence of POD.
2. Diagnostic Performance: To calculate the sensitivity and specificity of these distinct preoperative eye movement parameters for predicting POD development.
3. Incremental Predictive Value: To evaluate whether adding preoperative eye-tracking measurements to a standard clinical prediction model (comprising age, sex, CCI, and MoCA scores) significantly improves the model's overall predictive performance and risk stratification capabilities.