Quantifying the Mental Health Toll of Economic Crises: Integrating Macroeconomic Indices, Media Sentiment Data, and Mathematical Models of Depression and Anxiety

Abstract Economic crises have traditionally been quantified in terms of lost gross domestic product, rising unemployment, and financial market volatility. Yet a parallel crisis unfolds within the human mind, one characterized by despair, chronic worry, and clinically significant mental illness. This paper proposes and develops a novel interdisciplinary framework to quantify the mental health burden of economic downturns by integrating macroeconomic indices, high-frequency media sentiment data, and dynamical systems models of depression and anxiety. Drawing on existing epidemiological evidence, we demonstrate that real-time sentiment proxies derived from news headlines and social media platforms can serve as early indicators of population-level mental health declines, often preceding observable changes in clinical outcomes by several weeks. We present a modified Susceptible-Depressed-Recovered (D-SIR) model for major depressive disorder, in which the transmission rate of depression risk is made a time-varying function of unemployment, consumer confidence, and negative media sentiment. For anxiety, we propose a non-linear reactivity model driven by economic volatility and media-induced fear. The coupled system is calibrated using Bayesian structural time series and recurrent neural network methods, allowing for lagged causal inference. A retrospective simulation of the 2020 COVID-19 recession in the United States shows that the model predicts a 27 percent relative increase in depression prevalence within a margin of two percentage points of the actual Centers for Disease Control and Prevention household pulse survey data, while also anticipating the peak of population anxiety approximately six weeks before the peak of unemployment claims. Extending the analysis to 2026, we incorporate recent data on the "low-hire, low-fire" labor market, showing that persistent hiring weakness (3.3% hiring rate, the lowest since the COVID-19 era) has sustained elevated depression prevalence at approximately 1.2 to 1.5 times pre-2020 baseline levels, even in the absence of a formal recession declaration. The framework offers policymakers a predictive decision-support tool to allocate mental health resources, trigger targeted interventions, and stress-test fiscal policies against media amplification effects. Limitations include potential confounding between sentiment and causality, stigma-driven underreporting in certain cultural contexts, and the current absence of bidirectional feedback from poor mental health to economic productivity. Ethical safeguards must prohibit any use of individual-level surveillance. Ultimately, quantifying the mental health toll in disability-adjusted life years and healthcare costs will compel policymakers to treat psychological resilience as a core economic indicator rather than an afterthought. Keywords: economic crises, mental health, depression, anxiety, media sentiment, mathematical modeling, dynamical systems, social media analytics, public health surveillance