A MATHEMATICAL MODEL FOR PREDICTION OF LVEF FROM ECG IN A THREE-POINT DETERIORATION STUDY OF CARDIAC HEALTH IN DIABETICS

Cardiac diseases and diabetes are the two major causes of mortality across the world. In India, the mortality rate is three times higher due to diabetes. Electrocardiogram (ECG) and 2D Echocardiogram (Echo) are the major noninvasive techniques used for diagnosis. The research has two major findings, namely the study of cardiac health deterioration using three-point ECG data and mathematical model for prediction of 2D Echo from ECG that will help rural people to assess cardiac health with a degree of correctness equivalent to 2D Echo using ECG signal. ECG data of 16 diabetic subjects and 15 normal subjects in the supine position for 3 times, with a 6 months interval between two recordings, for 10 to 15 minutes were collected. The 2D echo data at the time of first ECG recording was also collected for diabetics. The result shows deterioration in linear and nonlinear parameters of heart rate variability (HRV) values from first readings to third in diabetic subjects, but not in healthy subjects; which implies deterioration of cardiac health in diabetic subjects. The correlation between low frequency/high-frequency ratio i.e. (LF/HF) (the ECG parameter) and ECHO parameter i.e. left ventricle ejection factor (LVEF), in diabetic subjects is found. Our results show a high correlation (72.92%) between LF/HF with LVEF in diabetic subjects. Combining with 3 point study result implies that there is deterioration in LVEF (Echo) also. These highly correlated values are used to develop a mathematical model to predict LVEF from LF/HF using logistic regression models. We choose an optimum model to give the best goodness of fit. This results in a polynomial equation with order 6 with R squared value 97.8%. The model helps to estimate LVEF (Echo) from LF/HF (ECG), thus providing an efficient and cost-effective method for heart checkups in rural areas where 2D echocardiography facility is lacking and costs ten times more than ECG.