Biotrack: Wireless Bioelectric Signal Acquisition and Classification Using Machine Learning

This work provides a holistic approach to classifying bioelectric signals, namely electrocardiograms (ECG) and electromyograms (EMG), with machine learning in healthcare applications. In this system, data are simulated through the use of software coupled with hardware implementation for real-time monitoring and accurate signal analysis. MATLAB is used for signal preprocessing, feature extraction, and training of a machine learning model. Noise reduction, normalization, and removal of artifacts have been used in preprocessing to make sure that goodquality data has been extracted. The meaningful patterns for the classification are drawn out using the Fourier Transform, Wavelet Transform, and statistical methods. This study compares the classification performance among three machine learning algorithms: Support Vector Machines, Random Forest, and Neural Networks. Neural Networks achieve the best accuracy (94 %) and are the best fit for high-stakes medical diagnostics, while Random Forest provides balanced performance with robustness to noise and fast inference times, which makes it a good fit for resource-constrained systems. Hardware implementation is based on different sensors, including AD8232 for ECG and MyoWare for EMG. In addition to the sensors, microcontrollers such as Arduino, STM32, and Raspberry Pi were used. In order to achieve wireless communication for real-time data acquisition and transference, modules of real-time clock, ADC, and band pass filters were integrated. Thus, the work bridges the gap between simulation and practical application and demonstrates a reliable and efficient bioelectric signal monitoring system. Future developments include improving the efficiency of algorithms, increasing datasets, and incorporating cloud-based analytics for advanced healthcare diagnostics