📄 中文摘要
大学生面临诸多压力，导致焦虑和抑郁水平普遍较高。可穿戴技术能提供非侵入式生理传感器数据，可用于精神疾病的早期检测。然而，现有研究在心理测量工具的多样性、生理模态以及时间序列参数方面存在局限。本研究旨在通过收集学生心理与环境健康（StudentMEH）数据，填补这些空白。通过利用Fitbit设备收集到的心率、睡眠模式、活动水平等生理指标，结合学生自我报告的心理健康问卷结果，构建预测模型。模型将深入分析这些生理数据与焦虑、抑郁等心理健康状况之间的关联，识别潜在的生物标志物。研究将探索不同时间尺度的生理数据变化，例如日间活动量、夜间睡眠质量、心率变异性等，如何反映学生心理状态的波动。此外，将考虑新冠疫情期间特有的环境因素，如社交隔离、学习模式改变等，对学生心理健康的影响。通过机器学习算法，如支持向量机、深度学习网络等，训练分类器以区分有心理健康风险和无心理健康风险的学生群体。最终目标是开发一个早期预警系统，利用可穿戴设备数据，在学生心理健康问题恶化之前进行识别和干预，从而为大学提供更有效的心理健康支持策略。此方法能够实现大规模、持续的心理健康监测，降低传统筛查方法的成本和时间消耗，并有助于理解复杂环境因素在心理健康发展中的作用。
📄 English Summary
Student Mental Health Screening via Fitbit Data Collected During the COVID-19 Pandemic
College students frequently experience significant stressors, leading to elevated levels of anxiety and depression. Wearable technology offers unobtrusive sensor data valuable for the early detection of mental illness. However, current research is often constrained by the limited variety of psychological instruments administered, physiological modalities analyzed, and time series parameters considered. This research addresses these limitations by collecting the Student Mental and Environmental Health (StudentMEH) dataset. Physiological metrics such as heart rate, sleep patterns, and activity levels, gathered from Fitbit devices, are integrated with self-reported psychological health questionnaire results to develop predictive models. The models will meticulously analyze the associations between these physiological data and mental health conditions like anxiety and depression, aiming to identify potential biomarkers. The study will explore how physiological data variations across different time scales—such as daily activity volumes, nocturnal sleep quality, and heart rate variability—reflect fluctuations in students' psychological states. Furthermore, the impact of unique environmental factors during the COVID-19 pandemic, including social isolation and altered learning modalities, on student mental health will be considered. Machine learning algorithms, including support vector machines and deep learning networks, will be employed to train classifiers capable of distinguishing between students at risk for mental health issues and those without. The ultimate goal is to develop an early warning system that utilizes wearable device data for identifying and intervening in student mental health problems before they escalate, thereby providing universities with more effective mental health support strategies. This approach enables large-scale, continuous mental health monitoring, reduces the cost and time associated with traditional screening methods, and contributes to understanding the role of complex environmental factors in mental health development.