Statistical methodology: VI. Mathematical modeling of the electrocardiogram using factor analysis

The ECG is a 12-lead-vector system and is known to contain redundant information. Factor analysis (FA) is a statistical technique that improves measured data and eliminates redundancy by identifying a minimum number of factors accounting for variance in the data set.

To identify the minimum number of lead-vectors required to predict the 12-lead ECG.

A total of 104 ECGs were obtained from 24 normal men, 22 normal women, and 28 men and 30 women with variable pathologies. Each ECG lead was simultaneously acquired and digitized, resulting in a voltage-time data array stored for mathematical analysis. Each array was factor-analyzed to identify the minimum number of lead-vectors spanning the ECG data space. The 12-lead ECG was then predicted from this minimum lead-vector set. ANOVA was used to test for statistical significance between normal and pathologic data groups.

FA revealed that 3 lead-vectors accounted for 99.12%+/-0.92% (95% CI+/-0.18%) of the variance contained in the 12-lead ECG voltage-time data for all 104 cases. There were no statistically significant differences between men and women (99.25%+/-0.66% vs 98.98+/-1.11%; p=0.139). Statistically significant differences were noted between normal and acute myocardial infarction ECGs (99.5%+/-0.27% vs 98.66+/-1.25%; p=0.00003). The measured and predicted leads were almost identical. A 3-dimensional spatial ECG derived from the 3-lead-vector set resulted in variable curved surfaces that differed by pathology.

The 12-lead ECG can be derived from only 3 measured leads and graphed as a 3-D spatial ECG. This type of data processing may lead to instantaneous acquisition and may enhance the diagnostic capability of the ECG from routine bedside telemetry equipment.