GLOBEM

该数据集名为Globem，它是一个包含来自移动设备和可穿戴传感器独特用户数据的广泛集合。该数据集不仅包括每周进行的生态瞬时评估（EMA）调查，以捕捉参与者的心理健康状况，还涵盖了各种健康指标的被动收集传感器数据。此外，数据集还包括位置、手机使用情况、蓝牙、通话、身体活动和睡眠等特征，并且调查数据基于患者健康问卷4（PHQ-4）。最终测试数据集总共包含90个类别平衡的样本。任务是通过多传感器数据对抑郁和焦虑进行二分类。

This dataset, named Globem, is a comprehensive collection of unique user data sourced from mobile devices and wearable sensors. It not only includes weekly Ecological Momentary Assessment (EMA) surveys that capture the mental health status of participants, but also encompasses passively collected sensor data for various health metrics. Additionally, the dataset features a range of attributes including location, mobile usage, Bluetooth, call logs, physical activity, and sleep. Furthermore, the survey data is grounded in the Patient Health Questionnaire 4 (PHQ-4). The final test dataset contains a total of 90 class-balanced samples. The core task is to perform binary classification of depression and anxiety using multi-sensor data.

搜集汇总

背景与挑战

背景概述

GLOBEM是一个多年度纵向行为建模数据集和算法平台，包含497名用户4年间的移动/可穿戴传感数据，支持抑郁症检测等行为建模任务，并实现了多种传统ML和DL算法。平台提供标准化评估框架和算法扩展接口，旨在促进跨数据集泛化研究。