Students are among the most vulnerable populations during periods of crisis, including war, economic collapse, and pandemics. These events extend beyond academic disruption, significantly affecting students' emotional and social well-being. Mental health challenges such as anxiety, depression, and behavioural changes are commonly reported, particularly among youth living in conflict-affected areas or economically disadvantaged households. This review examines the consequences of crises on school-aged students across both local and global contexts. A structured search strategy was employed to retrieve peer-reviewed articles published between 2005 and 2024 from databases including PubMed, ERIC, Scopus, and Google Scholar. The selected studies were thematically categorized into three primary domains: pandemics, economic hardship, and war-related trauma. The review emphasizes the identification of common psychological outcomes, contributing factors, and resilience strategies implemented at the school and community levels. The findings highlight the urgent need for early interventions, trauma-informed pedagogical approaches, mental health support programs, coping strategies, and emotional regulation skills. By examining the interplay between crisis-induced stress and student support mechanisms, this review seeks to inform educators, policymakers, and practitioners in their efforts to foster resilience and promote academic recovery.

This study applies explainable machine learning to identify which vocal attributes in a lecturer’s speech influence students’ views of a lecturer’s charisma, a key contributor to teaching quality. It further explores whether vocal qualities differ between male and female lecturers and how students of different genders respond to these differences, offering insights into voice-related factors that influence the impact of educators. Speech segments from YouTube videos featuring 200 native-English lecturers were evaluated by 900 students using charisma rating scales. A set of attributes related to three primary prosodic dimensions of voice - pitch, rhythm, and loudness - was computed. A random forest classifier was employed to predict the charisma level based on the speech attributes and to list and rank the attributes that contributed most to the prediction. The findings revealed prominent vocal attributes that achieved higher charisma scores in the students' ratings. Same-gender evaluations of charisma were mainly based on pitch, while cross-gender evaluations rely mostly on loudness or rhythm. The automated, interpretable method provides a reliable and efficient way to measure vocal charisma in academic lecturers. It can be adapted to examine additional individual factors that influence the perception of a lecturer’s charismatic presence. It may also be integrated into practice-based tools, designed to support instructors in improving their presentation skills. Our research bridges the fields of applied psychology and computer science to contribute to the development of educational technology.