Prediction of left ventricular mass from the electrocardiogram

Area of Science:

• Cardiology
• Biomedical Engineering
• Medical Imaging

Background:

• Left ventricular (LV) mass index is a crucial indicator of cardiac health.
• Accurate assessment of LV mass and hypertrophy is essential for cardiovascular risk stratification.
• Echocardiography is the gold standard for measuring LV mass, but it is resource-intensive.

Purpose of the Study:

• To develop and evaluate electrocardiogram (ECG) based models for predicting echocardiographic left ventricular (LV) mass index.
• To create a logistic regression model for classifying LV hypertrophy using standard 12-lead ECG measurements.
• To assess the performance of these ECG-derived models in predicting cardiac structural changes.

Main Methods:

• Utilized multiple stepwise regression techniques to derive predictive ECG models from 12-lead ECG data.
• Analyzed data from 203 men and 252 women to establish correlations between ECG and echocardiographic LV mass index.
• Developed a logistic regression classifier for LV hypertrophy using specific ECG leads (e.g., R in aVL, T in V6, S in V1/I).

Main Results:

• Achieved moderate correlations between ECG and echocardiographic estimates of LV mass index (R²=0.58 for men, R²=0.42 for women).
• The logistic regression model for LV hypertrophy demonstrated moderate sensitivity (53.7%-63.4%) and high specificity (92.9%-94.9%).
• Identified specific ECG leads (R in aVL, T in V6, S in V1/I) as significant predictors for LV mass and hypertrophy.

Conclusions:

• ECG-based models show promising initial performance for estimating LV mass index and classifying LV hypertrophy.
• These multivariate estimators offer a non-invasive and potentially more accessible approach compared to echocardiography.
• Further studies are recommended to validate the clinical utility of these ECG models as cardiovascular risk predictors.