| Popis: |
Heart Attacks are the major cause of death in the world today, particularly in India. The need to predict this is a major necessity for improving the country's healthcare sector. Accurate and precise prediction of the heart disease mainly depends on Electrocardiogram (ECG) data and clinical data. These data's must be fed to a non linear disease prediction model. This non linear heart function monitoring module must be able to detect arrhythmias such as tachycardia, bradycardia, myocardial infarction, atrial, ventricular fibrillation, atrial ventricular flutters and PVC's. In this paper we have developed an efficient method to acquire the clinical and ECG data, so as to train the Artificial Neural Network to accurately diagnose the heart and predict abnormalities if any. The overall process can be categorized into three steps. Firstly, we acquire the ECG of the patient by standard 3 lead pre jelled electrodes. The acquired ECG is then processed, amplified and filtered to remove any noise captured during the acquisition stage. This analog data is now converted into digital format by A/D converter, mainly because of its uncertainty. Secondly we acquire 4–5 relevant clinical data's like mean arterial pressure (MAP), fasting blood sugar (FBS), heart rate (HR), cholesterol (CH), and age/gender. Finally we use these two data's i.e. ECG and clinical data to train the neural network for classifying the heart disease and to predict abnormalities in the heart or it's functioning. |
