Abstract：
Adolescence is a developmental period in which social interactions are critical for mental health. While the onset of COVID-19 significantly disrupted adolescents\' social environments and mental health, it remains unclear how adolescents have adapted to later stages of the pandemic. We harnessed a machine learning architecture of Long Short-Term Memory recurrent networks (LSTM) with gradient-based feature importance, to model the association among daily social interactions and depressive symptoms during three stages of the pandemic. A year before COVID-19, 148 adolescents reported social interactions and depressive symptoms, every day for 21 days. One hundred sixteen of these youths completed a 28-day diary after schools closed due to COVID-19. Seventy-nine of these youths and additional 116 new participants completed a 28-day diary approximately a year into the pandemic. Our results show that LSTM successfully predicted depressive symptoms from at least a week of social interactions for all three waves (r2 > .70). Our study shows the utility of using an analytic approach that can identify temporal and nonlinear pathways through which social interactions may confer risk for depression. Our unique analysis of the importance of input features enabled us to interpret the association between social interactions and depressive symptoms. Collectively, we observed a return to pre-pandemic patterns a year into the pandemic, with reduced gender and age differences during the pandemic closures. This pattern suggests that the system of social influences in adolescence was affected by COVID-19, and that this effect was attenuated in more chronic stages of the pandemic.
摘要：
青春期是社会交往对心理健康至关重要的发展时期。虽然COVID-19的发作显著扰乱了青少年的社会环境和心理健康，目前尚不清楚青少年如何适应大流行的后期阶段.我们利用了具有基于梯度的特征重要性的长短期记忆循环网络（LSTM）的机器学习架构，在大流行的三个阶段，建立日常社交互动和抑郁症状之间的关联模型。在COVID-19前一年，148名青少年报告了社会交往和抑郁症状，每天21天。这些年轻人中有116人在学校因COVID-19关闭后完成了为期28天的日记。这些年轻人中有79名和另外116名新参与者在大流行大约一年后完成了为期28天的日记。我们的结果表明，LSTM成功地从至少一周的社交互动中预测了所有三波的抑郁症状（r2>.70）。我们的研究表明，使用分析方法可以识别时间和非线性途径，通过这些途径，社会互动可能会带来抑郁症的风险。我们对输入特征重要性的独特分析使我们能够解释社交互动与抑郁症状之间的关联。总的来说，我们观察到大流行一年后恢复到大流行前的模式，大流行关闭期间性别和年龄差异减少。这种模式表明，青春期的社会影响系统受到COVID-19的影响，这种影响在大流行的更慢性阶段减弱。
