AffectStream: Kafka-based real-time affect monitoring system using wearable sensors

Real-time affect monitoring is essential for personalized and adaptive applications in fields like education, healthcare, and customer service. However, existing systems often struggle with scalability and low-latency requirements for processing high-frequency sensor data. To address these challenges, we propose AffectStream , a Kafka-based real-time affect monitoring system that processes wearable sensor data through a cloud-based pub/sub architecture to the applications. AffectStream ensures scalability, fault tolerance, and personalized emotional state analysis. Its robust performance is demonstrated through trace-based evaluations using three public datasets (i.e., WESAD, AMIGOS, and GalaxyPPG). This open-source framework advances real-time emotion recognition, paving the way for large-scale affective computing applications. • AffectStream enables real-time affect monitoring using wearable sensor data. • Kafka-based pub/sub architecture ensures scalability and low-latency processing. • Personalized affect classification is performed in real time via cloud-based computing infrastructure. • Trace-based evaluation using three public datasets (i.e., WESAD, AMIGOS, and GalaxyPPG) validates system performance. • Open-source framework facilitates emotion-aware applications in various domains.