| dc.description.abstract | With the advancement of technology and deep learning in medical healthcare field
the usage of wearable devices for real-time monitoring biomedical signal, ECG signal for
instance, and analysis of long-term diseases received more attention. However, the
existing ECG monitoring devices in hospitals required high cost to store data, lack of
portable. In this paper, a real-time electrocardiogram (ECG) monitoring system is
proposed. The device can be used to measure the ECG signal, predict the beat positions
and beat types, display result on the web and save data to the cloud via the Raspberry Pi 4
MCP3008 ADC, and the sensor AD8232. The AD8232 sensor is used to record the signal,
while the MCP3008 ADC is utilized to convert the analog signal to digital signal so that
the Raspberry can read it. Once recorded signal, the Raspberry preprocess signal and run
the trained deep learning model to predict beat position and beat type. After that, the
signal is shown on the website and save data to cloud. The training result was 96%
accuracy for AHA and 84% for NSTDB dataset. The time for recorded and display signal
on the website is delayed within microseconds. The delayed time for predicting the result
is about 1s. The proposed system have an acceptable accuracy and a real-time recording
time when testing on author, showing a great potential for a fully portable device. | en_US |
