Toward Wearable Intelligence: Psychophysiological Responses and Digital Solutions to Climate Change

Global climate change poses profound threats to human psychophysiology. Digital health technologies and artificial intelligence (AI) offer unprecedented opportunities to monitor affective and cognitive mechanisms, support adaptive behaviours, improve the efficacy and reduce the carbon footprint of healthcare services delivery. This thesis has integrated wearable sensing, signal processing, and AI techniques to investigate human responses to climate stressors and to develop digital tools aimed at strengthening clinical practice and promoting sustainable health systems. A virtual reality-based simulation of a wildfire was designed to address the climate action-value gap. This immersive exposure elicited robust psychophysiological engagement, evaluated with heart rate variability and electrodermal activity, and improved climate knowledge and pro-environmental intentions, particularly in younger male participants. To overcome inherent biases of such self-reported measures, implicit attitudes (i.e., automatic beliefs that are not accessible through conscious introspection) toward climate change were assessed with eye tracking related parameters. Although explicit and implicit attitudes were not correlated, nonlinear ocular dynamics revealed higher cognitive load when participants processed incongruent climate-related associations, suggesting autonomic markers of environmentally motivated decision conflict. This outcome underlines the possibility to better evaluate the efficacy of educational programs and campaigns. This thesis then covered the development of digital health tools aligned with sustainability goals. Specifically, voice analysis was explored as a low-carbon, scalable biomarker for remote clinical care. AI-based diagnostic models, trained on sustained vowel and speech recordings, demonstrated the feasibility of distinguishing two types of voice disorders. Feature interpretability was also addressed, highlighting gender-specific markers to guide medical decision making. Following this outcome, one study automated post-treatment voice monitoring implementing an analogous strategy, whereas a further investigation adapted the concept of complexity matching in acoustic analysis and introduced a nonlinear, multivariate and multiscale framework to detect incomplete recovery of voice quality. These researches aimed at supporting patients follow-up, to identify individuals in need for additional care, and reducing unnecessary in-person visits. Finally, a smartphone-based voice analysis was applied to genetic syndromes utterances, achieving promising classification performance and revealing distinctive articulatory and phonatory patterns to support their non-invasive screening. Collectively, this work advanced multimodal sensing and explainable AI for understanding psychophysiological responses to climate stress and developing voice-based digital health solutions. It underlined how immersive experiences, implicit-cognition paradigms, and accessible technologies may jointly promote sustainable behaviours and healthcare, enhance self-awareness, and empower clinicians in a resource-efficient future.