A novel electrocardiogram-based model for prediction of dementia-The Atherosclerosis Risk in Communities (ARIC) study

Area of Science:

• Cardiology
• Neurology
• Gerontology

Background:

• Dementia prediction models are crucial for early intervention.
• Existing models like CAIDE have limitations in accuracy and calibration.
• Electrocardiogram (ECG) data offers a potential new source for predictive modeling.

Purpose of the Study:

• To develop and validate an ECG-based model for predicting dementia.
• To compare the predictive performance of the ECG model against the CAIDE model.
• To assess the clinical feasibility of an ECG-based dementia prediction tool.

Main Methods:

• Utilized data from the Atherosclerosis Risk in Communities study.
• Developed parsimonious ECG models using Cox regression and backward selection.
• Compared the C-statistic and calibration of the ECG model with the CAIDE model at two time points (V4 and V5).

Main Results:

• The ECG-based model demonstrated superior predictive performance compared to the CAIDE model at both V4 (C-statistic 0.72 vs. 0.67) and V5 (C-statistic 0.70 vs. 0.64).
• The ECG model showed good calibration, whereas the CAIDE model was poorly calibrated.
• The developed ECG model incorporated only two or three ECG variables and age.

Conclusions:

• A novel ECG-based model provides superior discrimination for predicting dementia in middle-aged and older adults compared to the CAIDE model.
• The ECG model's reliance on easily obtainable variables facilitates straightforward clinical implementation.
• This approach offers a promising, accessible tool for early dementia risk assessment.