Explainable AI for Mental Health Screening in Bangladesh: A Socio-Behavioral and Psychometric Dataset

This dataset contains anonymized survey responses from 9,984 adults in Bangladesh, collected to investigate the socio-behavioral and psychometric factors associated with mental health. The data was gathered as part of the research paper titled, "Explainable AI for Mental Health Screening in Bangladesh: A Measurement-Aware, Fairness-Audited, and Reproducible Framework." The primary goal of this data collection was to build and validate a machine learning framework capable of predicting depression severity, thereby addressing the challenges of mental health screening (e.g., cultural stigma, lack of structured data) in low- and middle-income countries. The dataset comprises a wide range of variables, including: Sociodemographic Information: Age, gender, marital status, educational attainment, employment status, geographic location (division), and financial indicators. Mental Health & Well-being Scales: Complete responses to validated psychometric questionnaires, including the Patient Health Questionnaire-9 (PHQ-9) for depression, the Generalized Anxiety Disorder-7 (GAD-7) for anxiety, and the Perceived Stress Scale-10 (PSS-10). Adverse Childhood Experiences (ACEs): Data related to experiences of abuse, neglect, and household dysfunction. Behavioral Factors: Self-reported data on internet usage patterns (based on the Internet Addiction Test - IAT), sleep duration, physical activity levels, and daily screen time. The data is provided in a single Comma-Separated Values (.csv) file. This resource is intended for researchers, data scientists, and public health professionals interested in the social determinants of mental health, the development of predictive models, and the auditing of algorithmic fairness in healthcare.本数据集包含孟加拉国9984名成年人的匿名化调查问卷回复，旨在探究与心理健康相关的社会行为及心理测量学因素。该数据采集工作作为题为《孟加拉国心理健康筛查的可解释人工智能（Explainable AI）：具备测量感知性、公平性审核与可复现性的框架》的研究论文的一部分完成。 本次数据采集的核心目标为构建并验证可预测抑郁严重程度的机器学习框架，以此解决中低收入国家心理健康筛查面临的各类挑战（如文化污名化、结构化数据匮乏等）。 本数据集涵盖多类变量，具体包括： 1. 社会人口统计学信息：年龄、性别、婚姻状况、受教育程度、就业状态、地理位置（行政区）以及财务指标。 2. 心理健康与幸福感量表：经过验证的心理测量问卷完整作答结果，包括用于评估抑郁的患者健康问卷-9（Patient Health Questionnaire-9, PHQ-9）、用于评估焦虑的广泛性焦虑障碍量表-7（Generalized Anxiety Disorder-7, GAD-7）以及感知压力量表-10（Perceived Stress Scale-10, PSS-10）。 3. 童年不良经历（Adverse Childhood Experiences, ACEs）：与虐待、忽视及家庭功能失调相关的数据。 4. 行为因素：基于网络成瘾测验（Internet Addiction Test, IAT）的自我报告式互联网使用模式数据、睡眠时长、身体活动水平及每日屏幕使用时间。 本数据集以单个逗号分隔值（Comma-Separated Values, .csv）文件形式提供。该资源面向关注心理健康社会决定因素、预测模型开发以及医疗领域算法公平性审核的研究人员、数据科学家与公共卫生专业人员开放。