| Abstrakt: |
Internet of Things (IoT) devices play a crucial role in the real-time acquisition of photoplethysmography (PPG) signals, facilitating seamless data transmission to cloud-based platforms for analysis. Atrial fibrillation (AF), affecting approximately 1–2% of the global population, requires accurate detection methods due to its prevalence and health impact. This study employs IoT devices to capture PPG signals and implements comprehensive preprocessing steps, including windowing, filtering, and artifact removal, to extract relevant features for classification. We explored a broad range of machine learning (ML) and deep learning (DL) approaches. Our results demonstrate superior performance, achieving an accuracy of 97.7%, surpassing state-of-the-art methods, including those with FDA clearance. Key strengths of our proposal include the use of shortened 15-second traces and validation using publicly available datasets. This research advances the design of cost-effective IoT devices for AF detection by leveraging diverse ML and DL techniques to enhance classification accuracy and robustness. [ABSTRACT FROM AUTHOR] |
