Physical Activity Indexing on Human ECG - A Biomarker for Clinical Diagnosis
The physical activity of a patient is directly affecting the cardiac electrical conduction levels. In order to avoid the misinterpretation of cardiac signal variation due to physical activity of the patient and the variation due to abnormal conduction of heart muscle, a biomarker is required to indicate the physical activity on ECG signal. This activity indexing on ECG helps to the clinician or doctor for the accurate diagnosis. The aim of this paper is to classify the physical activity and to indexing the activity segment on ECG. In this paper, the design of proposed system consists of ECG sensor and 3-axis Accelerometer connecting to Atmel328 microcontroller is used to collect the ECG and Accelerometer signals. The acquired signals will be stored in the SD card connected to the microcontroller. The signals will be recorded from 33 healthy volunteers under different activities such as supine, sitting, bending, standing, walking and running condition, for the proposed analysis. The statistical and energy features are extracted from each segment of activity. The activity detection is done by using machine learning algorithms. Finally, the activity type will be marked on ECG signal.
Keywords: Classification, Distributed Denial of Service Physical Activity, cardiac conduction, Electrocardiogram (ECG), activity detection, machine learning algorithm.