Discovering the influence of personal features in psychological processes using Artificial Intelligence techniques: the case of COVID-19’s lockdown in Spain

In late 2019, a novel coronavirus (SARS-CoV-2) outbreak in Wuhan, China, triggered the global COVID-19 pandemic. Spain detected its first cases in January 2020, and by mid-March, infections had exceeded 5,000, prompting a nationwide lockdown on March 14. While necessary to curb the spread, such restrictions posed serious psychological and socioeconomic challenges, particularly for vulnerable populations. This study investigates the psychological impact of lockdown and the factors influencing mental health using Artificial Intelligence methods. Data were collected via an online questionnaire and analyzed through two workflows, each involving three stages: (1) psychological assessment-based categorization, with or without unsupervised learning techniques; (2) training of various Machine Learning classifiers to differentiate between groups; and (3) feature importance analysis to identify key variables linked to psychological states. The models showed strong performance, with accuracies often above 90%, particularly for Random Forest, Decision Trees, and Support Vector Machines. Sensitivity and specificity metrics confirmed consistent performance across different psychological outcomes, with the health-related subset yielding the most reliable results. Models predicting vulnerability also achieved over 90% accuracy, although performance dropped slightly when focusing on less vulnerable individuals using only environmental and economic features. Findings indicate that depression is primarily influenced by general health factors, while agoraphobia is more closely tied to living conditions. Economic status significantly impacts vulnerability classification, yet financial stability alone does not guarantee better psychological outcomes. These results underscore the importance of considering a broad range of factors when developing public health strategies to address mental health during crises.