Stress detection using a novel human opinion dynamics based machine learning method

Abstract

In today's interconnected world, understanding societal mental health is more crucial than ever. While traditional machine learning methods have attempted to quantify stress from social media data, they often fail to capture the complexity of human emotion. Existing binary classification models, like Naive Bayes and SVM, are fundamentally limited; they can only categorize stress as present or absent, ignoring the nuanced spectrum of human stress levels. Moreover, these methods completely overlook the profound influence of social interactions, a key driver of individual and collective well-being. ​This research introduces a paradigm shift with its Human Opinion Dynamics Optimization (HODO) method. By integrating insights from social dynamics, the HODO model moves beyond a simple binary classification. It's the first of its kind to not only detect varying stress levels—from reduced to heightened—but also to quantify how interactions between individuals impact their mental state. This nuanced, socially-aware approach provides a more accurate and holistic assessment of societal stress. The implications are significant, offering a powerful tool for policymakers, public health officials, and mental health professionals to better understand and address the mental health challenges posed by a competitive and ever-changing world.