Predicting model for the impact of Internet usage characteristics on suicidal ideation among vocational high school students
-
摘要:
目的 探究职业高中学生(简称职高生)网络使用特征与自杀意念的关联，为职高生自杀群体精准干预提供参考依据。 方法 2023年3月，采用整群随机抽样方法从武汉市、咸宁市3所职业高中抽取1 781名学生为研究对象，采用哥伦比亚-自杀严重程度评价量表、网络成瘾量表分别评估自杀意念和网络成瘾状况。采用LASSO回归模型筛选自杀意念的影响因素，采用梯度增强决策树算法XGBoost构建自杀意念预测模型，评估预测效能，通过计算SHAP值解释每个影响因素的贡献程度。 结果 职高生的自杀意念报告率为42.22%，网络成瘾报告率为26.39%。LASSO回归结果显示，年龄、性别、留守经历、父母关系、担任班干部、上网内容为学习、上网时间段(黎明、清晨、深夜)、网络成瘾、抑郁症状均是职高生自杀意念的影响因素(β值分别为-0.05，0.29，0.09，0.27，0.10，-0.01，0.09，0.05，0.24，0.28，0.78，P值均＜0.05)，预测模型的AUC值为0.75。基于SHAP的可视化结果表明，所有通过多变量分析确定的影响因素对模型预测均有正向贡献(SHAP值均>0)，其中抑郁症状、父母关系对自杀意念的影响最大(SHAP值分别为0.77，0.26)，贡献度较大的特征值之间共同作用可以提高预测概率。 结论 抑郁症状、父母关系、网络成瘾和上网时间段均是职高生自杀意念的重要风险因素，应进行有效干预降低职高生自杀风险。 Abstract:Objective To explore the association between the Internet usage characteristics and suicidal ideation among vocational high school students, so as to provide a theoretical basis for precise intervention of suicide among vocational high school students. Methods A total of 1 781 students were recruited from three vocational high schools in Wuhan and Xianning in March 2023 by using the cluster random sampling method. The Columbia-Suicide Severity Rating Scale and Revised Chen Internet Addiction Scale were used to measure suicidal ideation and Internet addiction, respectively. LASSO regression model was used to select influential factors related to suicidal ideation, and the gradient boosting decision tree algorithm XGBoost was used to develop prediction models and evaluate predictive performance. By calculating the SHAP values, the contribution of each influential factor was quantified. Results The prevalence of suicidal ideation among vocational high school students was 42.22% and prevalence of Internet addiction was 26.39%. LASSO regression results indicated that age, gender, experience of being left behind, parental relationship, holding a class cadre position, using the Internet for learning, Internet use during dawn, morning and late night, Internet addiction, and depressive symptoms were all the influential factors of suicidal ideation among vocational high school students (β=-0.05, 0.29, 0.09, 0.27, 0.10, -0.01, 0.09, 0.05, 0.24, 0.28, 0.78, all P < 0.05). The AUC of the prediction model was 0.75. The results based on SHAP values indicated that all influential factors identified through multivariate analysis contributed positively to the model predictions (SHAP>0). Among these, depressive symptoms and parental relationship had the greatest impact on suicidal ideation (SHAP=0.77, 0.26), and the joint effect of features with higher contribution could improve the prediction probability. Conclusions Depressive symptoms, parental relationships, Internet addiction, and time of Internet use are most important risk factors of suicidal behaviors for vocational high school students. Thus, effective interventions should be conducted to reduce their suicidal ideation. -
Key words:
- Internet /
- Suicide /
- Mental health /
- Regression analysis /
- Students
1) 利益冲突声明 所有作者声明无利益冲突。 -
图 1 职高生自杀意念SHAP值分析
注：A模型基于SHAP的特征重要性分析，B为SHAP特征摘要分析。
Figure 1. SHAP value analysis of suicidal ideation among vocational high school students
图 2 职高生抑郁症状与父母关系和深夜上网的部分特征依赖分析
注：父母关系1~5分别指很融洽、比较融洽、一般、较紧张、很紧张，抑郁症状0~2分别指无、可能有、肯定有。
Figure 2. Partial feature dependency of depression symptoms, parental relationship and late-night Internet usage among vocational high school students
表 1 不同组别职高生自杀意念报告率比较
Table 1. Comparison of detection rate of suicidal ideation among different groups of vocational high school students
组别 选项 人数 自杀意念报告人数 χ2值 P值 组别 选项 人数 自杀意念报告人数 χ2值 P值 性别 男 995 359(36.08) 34.31 < 0.01 每天平均上网时长/h < 3 604 226(37.42) 17.40 < 0.01 女 786 393(50.00) 3~ < 4 406 168(41.38) 留守经历 否 1 296 499(38.50) 26.45 < 0.01 4~ < 6 415 176(42.41) 是 485 253(52.16) ≥6 356 182(51.12) 父母关系 很融洽 578 167(28.89) 114.94 < 0.01 上网学习 否 564 271(48.05) 11.14 < 0.01 比较融洽 599 240(40.07) 是 1 217 481(39.52) 一般 381 194(50.92) 上网社交聊天 否 119 43(36.13) 1.68 0.20 较紧张 162 112(69.14) 是 1 662 709(42.66) 很紧张 61 39(63.93) 上网看视频 否 174 64(36.78) 2.10 0.15 教育方式 放任自由 137 74(54.01) 56.72 < 0.01 是 1 607 688(42.81) 比较民主 1 440 548(38.06) 上网听音乐 否 345 133(38.55) 2.18 0.14 比较强制 204 130(63.73) 是 1 436 619(43.11) 是否重点班 否 1 483 623(82.85) 0.12 0.73 上网浏览各种APP(知乎、小红书、贴吧、微博等) 否 1 021 421(41.23) 0.87 0.35 是 298 129(17.15) 是 760 331(43.55) 是否担任班干部 否 1 336 540(40.42) 6.84 0.01 上网看小说 否 1137 475(41.78) 0.21 0.65 是 445 212(47.64) 是 644 277(43.01) 成绩班级排名 前10% 129 55(7.31) 10.69 0.06 上网干其他事情 否 1 626 687(42.25) 0.01 1.00 前20% 228 95(12.63) 是 155 65(41.94) 前30% 218 94(12.50) 拂晓(00:00—02:59)上网 否 825 289(35.03) 32.05 < 0.01 前40% 168 58(7.71) 是 956 463(48.43) 前50% 434 169(22.47) 黎明(03:00—05:59)上网 否 1 322 500(37.82) 40.05 < 0.01 50%以后 604 281(37.37) 是 459 252(54.90) 抑郁症状 无 896 205(22.88) 321.99 < 0.01 清晨(06:00—08:59)上网 否 586 202(34.47) 21.05 < 0.01 可能有 243 106(43.62) 是 1 195 550(46.03) 肯定有 642 441(68.69) 上午(09:00—11:59)上网 否 484 192(39.67) 1.64 0.20 首次使用网络方式 手机 1 262 513(40.65) 2.45 0.08 是 1 297 560(43.18) 电脑 385 174(45.19) 中午(12:00—14:59)上网 否 14 41(29.08) 10.27 < 0.01 平板 126 63(50.00) 是 1 640 711(43.35) 其他 8 2(25.00) 下午(15:00—17:59)上网 否 292 106(36.30) 4.74 0.03 首次使用网络地点 家里 1 618 686(42.40) 0.55* 是 1 489 646(43.38) 网吧 41 14(34.15) 傍晚(18:00—20:59)上网 否 162 52(32.10) 7.04 0.01 学校 30 14(46.67) 是 1 619 700(43.38) 亲朋好友 86 34(39.53) 深夜(21:00—23:59)上网 否 335 89(26.57) 40.67 < 0.01 其他 6 4(66.67) 是 1 446 663(45.85) 上网打游戏 否 348 133(38.22) 2.64 0.10 网络成瘾 否 1 311 470(35.85) 81.72 < 0.01 是 1 433 619(43.20) 是 470 282(60.00) 注：()内数字为报告率/%，*为Fisher检验。 
下载: 导出CSV
-
[1] World Health Organization. Suicide prevention[EB/OL]. (2021-12-06)[2024-08-14]. https://www.who.int/mental_health/suicide-prevention/en/ .[2] World Health Organization. Adolescent mental health[EB/OL]. (2023-06-08)[2024-08-14]. https://www.who.int/news-room/fact-sheets/detail/adolescent-mental-health .[3] 中华人民共和国国家卫生健康委员会. 2022中国卫生健康统计年鉴[M]. 北京: 中国协和医科大学出版社, 2022: 286-302.National Health Commission of the PRC. China health statistics yearbook in 2022[M]. Beijing: Peking Union Medical College Press, 2022: 286-302. (in Chinese) [4] ZHU Y, LI Y, XU X. Suicidal ideation and suicide attempts in psychiatric patients during the COVID-19: a systematic review and Meta-analysis[J]. Psychiatry Res, 2022, 317: 114837. doi: 10.1016/j.psychres.2022.114837 [5] YANG Y, ZHANG E L, LIU Y, et al. Network analysis of suicidality and Internet addiction symptoms among Chinese primary and secondary school students[J]. J Affect Disord, 2023, 339: 145-152. doi: 10.1016/j.jad.2023.07.030 [6] GROCOTT L R, MAIR A, GALIONE J N, et al. Days with and without self-injurious thoughts and behaviors: impact of childhood maltreatment on adolescent online social networking[J]. J Adolesc, 2022, 94(5): 748-762. doi: 10.1002/jad.12060 [7] ZHANG Y, WANG D, MA Z, et al. Problematic Internet use and suicide ideation among Chinese adolescents: the indirect effects of insomnia, nightmares, and social jetlag[J]. J Affect Disord, 2024, 344: 347-355. doi: 10.1016/j.jad.2023.10.081 [8] HUANG Y, XU L, MEI Y, et al. Problematic Internet use and the risk of suicide ideation in Chinese adolescents: a cross-sectional analysis[J]. Psychiatry Res, 2020, 290: 112963. doi: 10.1016/j.psychres.2020.11296304122 [9] 廖娟娟, 郑亚楠, 黄亮明, 等. PHQ-2-C、CES-D和BDI-Ⅱ-C在筛查青少年抑郁障碍中的应用研究[J]. 国际精神病学杂志, 2017, 44(1): 23-25, 36.LIAO J J, ZHENG Y N, HUANG L M, et al. Study of PHQ-2-C, CES-D and BDI-Ⅱ-C in the application of screening of adolescents with depressive disorder[J]. J Int Psychiatry, 2017, 44(1): 23-25, 36. (in Chinese) [10] MAK K K, LAI C M, KO C H, et al. Psychometric properties of the Revised Chen Internet Addiction Scale (CIAS-R) in Chinese adolescents[J]. J Abnorm Child Psychol, 2014, 42(7): 1237-1245. doi: 10.1007/s10802-014-9851-3 [11] BECK A T, BECK R, KOVACS M. Classification of suicidal behaviors: Ⅰ. quantifying intent and medical lethality[J]. Am J Psychiatry, 1975, 132(3): 285-287. doi: 10.1176/ajp.132.3.285 [12] 汪妍, 彭畅, 程君菡, 等. 中国5省在校中学生自杀行为流行状况及其与抑郁、焦虑关联[J]. 中国公共卫生, 2023, 39(10): 1225-1231.WANG Y, PENG C, CHENG J H, et al. Prevalence of and impact of depression and anxiety on suicidal behaviors among high school students: a cross-sectional survey in five provinces, China[J]. Chin J Public Health, 2023, 39(10): 1225-1231. (in Chinese) [13] 邹广顺, 吕军城, 乔晓伟. 中国中学生自杀意念检出率的Meta分析[J]. 中国心理卫生杂志, 2021, 35(8): 643-650.ZOU G S, LV J C, QIAO X W. A Meta-analysis of detection rate of suicidal ideation in middle school students in China[J]. Chin Ment Health J, 2021, 35(8): 643-650. (in Chinese) [14] LYONS A, HILL A O, MCNAIR R, et al. Demographic and psychosocial factors associated with recent suicidal ideation and suicide attempts among lesbian, gay, bisexual, pansexual, queer, and asexual (LGBQ) people in Australia: correlates of suicidality among LGBQ Australians[J]. J Affect Disord, 2022, 296: 522-531. doi: 10.1016/j.jad.2021.09.105 [15] PARK C H K, LEE J W, LEE S Y, et al. Suicide risk factors across suicidal ideators, single suicide attempters, and multiple suicide attempters[J]. J Psychiatr Res, 2020, 131: 1-8. doi: 10.1016/j.jpsychires.2020.08.018 [16] FELLMETH G, ROSE-CLARKE K, ZHAO C, et al. Health impacts of parental migration on left-behind children and adolescents: a systematic review and Meta-analysis[J]. Lancet (London, England), 2018, 392(10164): 2567-2582. doi: 10.1016/S0140-6736(18)32558-3 [17] 杨青, 宋薇, 粟芳. 有自杀意念的医学生儿童时期留守记忆的质性研究[J]. 心理月刊, 2023, 18(9): 69-71, 86.YANG Q, SONG W, SU F. A qualitative study on the left-behind memories of medical students with suicidal ideation in childhood[J]. Psychol Mon, 2023, 18(9): 69-71, 86. (in Chinese) [18] YANG Q, HU Y Q, ZENG Z H, et al. The relationship of family functioning and suicidal ideation among adolescents: the mediating role of defeat and the moderating role of meaning in life[J]. Int J Environ Res Public Health, 2022, 19(23): 15895. doi: 10.3390/ijerph192315895 [19] 王存同, 韩婷婷, 张杰. 大学生自杀意念的外在保护因素分析[J]. 中国心理卫生杂志, 2022, 36(3): 243-247.WANG C T, HAN T T, ZHANG J. External protective factors of suicidal ideation in college students[J]. Chin Ment Health J, 2022, 36(3): 243-247. (in Chinese) [20] 罗逸浩, 黄艳苹, 苏斌原. 高校研究生抑郁、焦虑、孤独感和自杀意念的关系[J]. 中国健康心理学杂志, 2021, 29(9): 1412-1415.LUO Y H, HUANG Y P, SU B Y. Relationship among depression, anxiety, loneliness, and suicidal ideation of graduate students in colleges and universities[J]. China J Health Psychol, 2021, 29(9): 1412-1415. (in Chinese) [21] 徐莹, 廖若男, 邱霜, 等. 中学生昼夜节律情绪调节与抑郁症状的相关性[J]. 中国学校卫生, 2022, 43(5): 722-726.XU Y, LIAO R N, QIU S, et al. Correlation between circadian rhythm, emotion regulation and depressive symptoms among adolescents[J]. Chin J Sch Health, 2022, 43(5): 722-726. (in Chinese) [22] 郭策, 段晓楠, 来亚平, 等. 青少年抑郁症状与清晨觉醒唾液皮质醇水平的相关性[J]. 中国学校卫生, 2018, 39(4): 540-542, 545. doi: 10.16835/j.cnki.1000-9817.2018.04.017GUO C, DUAN X N, LAI Y P, et al. Relationship between depressive symptoms and salivary awakening cortisol concentration in adolescents[J]. Chin J Sch Health, 2018, 39(4): 540-542, 545. (in Chinese) doi: 10.16835/j.cnki.1000-9817.2018.04.017 [23] DONG G, POTENZA M N. A cognitive-behavioral model of Internet gaming disorder: theoretical underpinnings and clinical implications[J]. J Psychiatr Res, 2014, 58: 7-11. doi: 10.1016/j.jpsychires.2014.07.005 -
点击查看大图
图(2) / 表(1)
计量
- 文章访问数: 11
- HTML全文浏览量: 7
- PDF下载量: 3
- 被引次数: 0
