Veneto Dynamic Observatory Proposal of Artificial Intelligence Systems for Digital Mental Health in Emergency Analysis.

This research aims to propose AI systems for analyzing the configuration of emergencies and their impacts on digital mental health, with particular reference to the Veneto Region community. The research project has been funded through a grant from Kalky s.r.l., in relation to testing AI systems for natural language analysis and applying the Methodology for the Analysis of Computerized Text Data (MADIT). Emergency management has been a widely discussed topic in recent years, especially considering the various types of emergencies related to climate change, natural disasters, health, and armed conflicts— all types of emergencies that have had and will continue to have a significant impact on human interactions. The recent Covid-19 pandemic highlighted a strong peak of uncertainty, triggering a global emergency that raised concerns not only in terms of health, but also socially, economically, and environmentally. In general, these emergencies have shown politicians and decision-makers, investors, and citizens that even individual natural events such as pandemics, if underestimated, can profoundly challenge and alter our lives, society, governance, and health, on an unprecedented scale. The research goal will be achieved through the analysis of data about different emergencies, collected from social media and from the administration of self-reports with multiple choice and open-ended questions to selected participant groups. The project involves data analysis using statistical tools for both quantitative and qualitative data (statistical tools for semi-automated text analysis and experimentation with a BERT-based model and LLM), and comparison with the existing literature. The application and use of this data collection will provide a broad and in-depth overview of the psychological effects generated by emergencies at various levels, and the impact these have on digital mental health. The AI systems employed for analyzing all the data collected will enable potential policymakers to: have access to indicators related to the psychological needs and make effective and efficient decisions for the benefit of the public, based on the specific community structure of the Region. The research work focused particularly on two types of emergencies: pandemic (Covid-19) and climatic. Six studies were conducted: two of them explored the impact of emergencies on mental health in relation to specific psychological constructs and the artificial intelligence systems applicable to such topic, and four explored the methodological issues. Subsequently, the developments in the methodological and applicative dimension were explored in relation to the classification system chosen by the theme-bound project, in the field of AI systems. For this reason, the experiments were carried out with 3 AI systems, improving the performance of previous research and implementing innovative prompts dedicated to the methodology adopted and studied: BERT, LLaMA, and Gemini.