HYBRID FUZZY LOGIC AND DEEP LEARNING FRAMEWORK FOR MOOD PREDICTION: A CROSS-DATASET STUDY

Abstract

Mood is the representation of identifying and interpreting an individual's emotional state of mind. Changes in emotional states manifest through alterations in communication frequency, typing patterns, app usage duration, physical activity levels, social interaction preferences, and even subtle variations in gait or posture captured by wearable devices. In recent years, advances in technology have made it possible to infer moods accurately in real time, providing crucial insights that enhance human-machine interaction, healthcare, marketing, and social connectivity. By aggregating and analyzing these diverse data streams using machine learning algorithms and pattern recognition techniques, Internet of behavior (IoB) systems can identify mood states  through the analysis of behavioral and physiological signals. To address this concerns, mood detection and IoB form a powerful synergy. IoB facilitates continuous monitoring and contextual analysis of behavior, enabling dynamic and precise mood detection. Mood are derived into different categories based on IoB parameter applied on E-DAIC and Reddit dataset. This classification is validated using PHQ-score and fuzzy logic.  Random forest achieved 71.43% remarkable accuracy, 68.03% precision score and classified mood into depressed with 96% recall, 92.13% precision and 94.12% F1-score. These predictions give early signs of emotional and psychological issues.   This combination has the ability to transform personalization efforts, psychological interventions, marketing campaigns, work environments, and other aspects through the development of an emotional system that can respond with empathy.