ARM-Based AI Module for Ectopic beat detection

This paper presented an ARM-based intelligent healthcare monitoring system for real-time ECG arrhythmia classification using deep learning and Internet of Things (IoT) technologies. The proposed system successfully integrated ECG signal acquisition, physiological parameter monitoring, signal preprocessing, Continuous Wavelet Transform (CWT)-based feature extraction, Convolutional Neural Network (CNN)-based arrhythmia classification, and IoT-enabled remote healthcare monitoring into a unified embedded healthcare platform. The Raspberry Pi ARM-based embedded system effectively interfaced with multiple biomedical sensors including ECG electrodes, MAX30102 pulse oximeter sensor, DHT11 temperature sensor, and MPU6050 accelerometer sensor for continuous physiological monitoring. Signal preprocessing techniques such as filtering, normalization, smoothing, and R-peak detection improved ECG signal quality and enabled accurate heartbeat segmentation for arrhythmia analysis. The application of Continuous Wavelet Transform provided detailed time-frequency representations of ECG signals, improving feature extraction capability for deep learning analysis. The CNN model successfully classified ECG signals into different arrhythmia categories such as Normal, Supraventricular, Ventricular, Fusion, and Unknown. Experimental results demonstrated that the integration of CWT and CNN significantly improved classification efficiency and enabled reliable real-time arrhythmia detection. The Flask-based IoT dashboard successfully enabled remote healthcare monitoring by displaying ECG waveforms, physiological parameters, arrhythmia classification results, historical health records, and alert notifications through internet connectivity. The system also generated emergency alerts during abnormal physiological conditions and fall events, improving patient safety and emergency healthcare response capability. The proposed system provides several advantages including portability, low implementation cost, intelligent ECG analysis, remote healthcare accessibility, and real-time physiological monitoring. The integration of embedded systems, biomedical signal processing, deep learning, and IoT communication demonstrates an efficient approach for next-generation smart healthcare applications suitable for telemedicine, elderly care, home healthcare, and continuous patient supervision. Future enhancements may include implementation of advanced deep learning architectures, cloud-based healthcare analytics, wearable healthcare integration, improved cybersecurity mechanisms, and large-scale clinical validation for enhanced diagnostic performance and healthcare reliability.

[1]    Xie Y., Wang H., & Liu Z. (2024). “ARM-Based AI Module for Ectopic Beat Classification Using Lightweight CNN.” IEEE Access, 12, 11234-11245. [2]    Morshed M., Rahman S., & Karim A. (2023). “On-Chip AI Classifier for Real-Time Arrhythmia Detection in Wearable ECG Systems.” Sensors, 23(14), 1-15. [3]  Zhang Y., Chen X., & Li H. (2021). “Lightweight Convolutional Neural Networks for ECG Arrhythmia Classification on Wearable Devices.” Biomedical Signal Processing and Control, 68, 102110. [4]  Raj P., Kumar S., & Reddy V. (2021). “Energy Efficient Embedded CNN for Real-Time ECG Monitoring Using ARM Processors.” International Journal of Embedded Systems, 14(2), 88-96. [5]  Neumann D., & Tejedor J. (2020). “CNN-Based ECG Signal Processing for ARM Cortex-M4 Embedded Systems.” IEEE Embedded Systems Letters, 12(4), 95-99. [6]  Rajpurkar P., Hannun A. Y., Haghpanahi M., Bourn C., & Ng A. Y. (2019). “Cardiologist-Level Arrhythmia Detection with Convolutional Neural Networks.” Nature Medicine, 25(1), 65- 69. [7]  Hannun A. Y., Rajpurkar P., Haghpanahi M., Tison G. H., Bourn C., Turakhia M. P., & Ng A. Y. (2019). “Deep Neural Networks for ECG Classification.” Nature Medicine, 25(1), 65- 69. [8]  Kiranyaz S., Ince T., & Gabbouj M. (2016). “Real-Time Patient-Specific ECG Classification by 1-D Convolutional Neural Networks.” IEEE Transactions on Biomedical Engineering, 63(3), 664-675. [9]  LeCun Y., Bengio Y., & Hinton G. (2015). “Deep Learning.” Nature, 521(7553), 436-444.Ince T., Kiranyaz S., & Gabbouj M. (2012). “A Generic and Robust System for Automated Patient- Specific Classification of ECG Signals.” IEEE Transactions on Biomedical Engineering, 56(5), 1415-1426. 10. Pan J., & Tompkins W. J. (1985). “A Real-Time QRS Detection Algorithm.” IEEE Transactions on Biomedical Engineering, BME-32(3), 230-236.