PPG-Based Real-Time Heart Rhythm Analysis Using Artificial Intelligence

This study presents a real-time heart rhythm analysis system based on photoplethysmography (PPG) signals using artificial intelligence techniques. PPG data were collected through low-cost, non-invasive optical sensors and processed in real time on an edge-computing platform. After signal preprocessing steps including band-pass filtering, baseline correction, and motion artifact reduction, time-domain, frequency-domain, morphological, and nonlinear features were extracted. Both machine learning and deep learning models were evaluated, including ensemble-based LightGBM and Long Short-Term Memory (LSTM) networks. Experimental results show that lightweight ensemble models achieve high classification accuracy with inference latency below 100 ms on edge devices, while deep learning models demonstrate higher robustness against noisy and artifact-contaminated signals. The proposed approach provides a reliable and low-latency solution for real-time heart rhythm monitoring and clinical decision support systems.