为了对学生异常行为的早期感知及校园行为时序建模,提出一种异常行为敏感的学生行为时序建模及心理健康预测(student behavioral temporal modeling sensitive to abnormal behavior for mental health prediction, SBTM-SABMHP)方法,...为了对学生异常行为的早期感知及校园行为时序建模,提出一种异常行为敏感的学生行为时序建模及心理健康预测(student behavioral temporal modeling sensitive to abnormal behavior for mental health prediction, SBTM-SABMHP)方法,利用移动设备收集的加速器、声音传感器及移动热点(wireless fidelity, WI-FI)等多种行为感知数据,构建异质信息网络,对学生当前行为模式进行建模。同时,为实现对学生历史行为时序数据的建模,建立了基于注意力机制的异常行为敏感的门控模块,有效融合学生长短期行为,并对学生行为时序建模,实现心理健康预测。在公共数据集StudentLife上对所提出的模型进行了对比分析实验。实验结果表明,与多种学生心理健康预测基线方法相比,该方法在4个评价指标上都取得了最佳性能,证明了该模型在学生心理健康预测任务上的有效性。展开更多
Objectives:First-year college students had exposure to unhealthy lifestyle behaviors that correlate with a high prevalence of anxiety and depression.Regarding to the modifiable lifestyle behaviors factors,this study i...Objectives:First-year college students had exposure to unhealthy lifestyle behaviors that correlate with a high prevalence of anxiety and depression.Regarding to the modifiable lifestyle behaviors factors,this study investigated the prevalence and correlation of multiple lifestyle behaviors,anxiety and depression in a sample of Chinese first-year college students.Methods:Cross-sectional data were extracted from Residents eHealth app of health lifestyle behaviors survey from September to October 2019.Anxiety,depression,eating regular meals,consumption of snacks in-between meals,consumption of fruit,dessert and sugar-sweetened beverages,smoking and secondhand smoke exposure,consuming alcohol,physical activity,sedentary time were assessed by self-report.Socio-demographic including age,gender,education,family income,religion,and health condition were captured.Logistic regression was used to explore the association of multiple lifestyle behaviors,anxiety and depression.Results:Totally 1,017 participants were included in the study.The prevalence of anxiety and depression(from mild to severe) were 40.3% and 45.3%,respectively.In multivariable analyses,religion (believe in Buddhism,OR =2.438,95%CI:1.097-5.421;believe in Christian,OR =5.886,95%CI:1.604-21.597),gender (Female,OR =1.405,95%CI:1.001-1.971),secondhand smoke exposure (OR =1.089,95%CI:1.001-1.184),and eating regular meals (OR =0.513,95%C1:0.346-0.759) were associated with anxiety.Family income (OR =0.732,95%CI:0.596-0.898),eating regular meals (OR =0.641,95%CI:0.415-0.990),frequency of breakfast (OR =0.813,95%CI:0.690-0.959),with a chronic disease (OR =1.902,95%CI:1.335-2.712),and consumption of nocturnal snack (OR =1.337,95%CI:1.108-1.612) were associated with depression.Conclusions:These results highlighted the need for early lifestyle behavior intervention,especially modifying diet patterns considering the background of religion,health condition,and social-economic status in first-year college students to improve their mental health.展开更多
文摘为了对学生异常行为的早期感知及校园行为时序建模,提出一种异常行为敏感的学生行为时序建模及心理健康预测(student behavioral temporal modeling sensitive to abnormal behavior for mental health prediction, SBTM-SABMHP)方法,利用移动设备收集的加速器、声音传感器及移动热点(wireless fidelity, WI-FI)等多种行为感知数据,构建异质信息网络,对学生当前行为模式进行建模。同时,为实现对学生历史行为时序数据的建模,建立了基于注意力机制的异常行为敏感的门控模块,有效融合学生长短期行为,并对学生行为时序建模,实现心理健康预测。在公共数据集StudentLife上对所提出的模型进行了对比分析实验。实验结果表明,与多种学生心理健康预测基线方法相比,该方法在4个评价指标上都取得了最佳性能,证明了该模型在学生心理健康预测任务上的有效性。
基金This research did not receive any specific grant from funding agencies in the public,commercial,or not-for-profit sectors.
文摘Objectives:First-year college students had exposure to unhealthy lifestyle behaviors that correlate with a high prevalence of anxiety and depression.Regarding to the modifiable lifestyle behaviors factors,this study investigated the prevalence and correlation of multiple lifestyle behaviors,anxiety and depression in a sample of Chinese first-year college students.Methods:Cross-sectional data were extracted from Residents eHealth app of health lifestyle behaviors survey from September to October 2019.Anxiety,depression,eating regular meals,consumption of snacks in-between meals,consumption of fruit,dessert and sugar-sweetened beverages,smoking and secondhand smoke exposure,consuming alcohol,physical activity,sedentary time were assessed by self-report.Socio-demographic including age,gender,education,family income,religion,and health condition were captured.Logistic regression was used to explore the association of multiple lifestyle behaviors,anxiety and depression.Results:Totally 1,017 participants were included in the study.The prevalence of anxiety and depression(from mild to severe) were 40.3% and 45.3%,respectively.In multivariable analyses,religion (believe in Buddhism,OR =2.438,95%CI:1.097-5.421;believe in Christian,OR =5.886,95%CI:1.604-21.597),gender (Female,OR =1.405,95%CI:1.001-1.971),secondhand smoke exposure (OR =1.089,95%CI:1.001-1.184),and eating regular meals (OR =0.513,95%C1:0.346-0.759) were associated with anxiety.Family income (OR =0.732,95%CI:0.596-0.898),eating regular meals (OR =0.641,95%CI:0.415-0.990),frequency of breakfast (OR =0.813,95%CI:0.690-0.959),with a chronic disease (OR =1.902,95%CI:1.335-2.712),and consumption of nocturnal snack (OR =1.337,95%CI:1.108-1.612) were associated with depression.Conclusions:These results highlighted the need for early lifestyle behavior intervention,especially modifying diet patterns considering the background of religion,health condition,and social-economic status in first-year college students to improve their mental health.
