Implementation of Artificial Neural Network for Real-Time Blood Pressure Estimation using ECG Signal

Hypertension is one of the serious health problems because patients will realize when the disease becomes more severe or complications arise. Many studies have used a combination of electrocardiogram (ECG) signals and photoplethysmography (PPG) signals with various methods. Therefore, it is inefficient because it requires more sources of information that requires more sensors and time consuming. To solve this problem, we proposed real time blood pressure (BP) estimation using ECG signal and age based on Artificial Neural Network (ANN). ECG signal is extracted using AD8232 module and processed using Arduino Nano. Then, the features are selected using curve fitting and used for model training. A total of 56 samples were taken from healthy subjects aged 19-58 years. The results show high correlation between blood pressure and ECG signal, which is shown in high coefficient correlation. Our method shows promising results with Mean Absolute Error (MAE) values on training performances of 2.4 mmHg for systolic blood pressure (SBP) and 2.2 mmHg for diastolic blood pressure (DBP). The MAE values for SBP and DBP prediction using 15 separated test data are 2.8 mmHg and 2.9 mmHg, respectively. Experimental results show that our proposed method can estimate blood BP in real time with good accuracy.