States and localities nationwide took unprecedented steps to reduce public health threats posed by COVID-19, including school closures affecting over 50 million youth. The pandemic also caused families extreme financial hardship, sudden unemployment, and distress. This combination of collective trauma, social isolation, and economic recession drastically increased the risk for adolescent major depression (MD), which was already the leading cause of disability in youth. However, youth MD treatments faced problems of potency and accessibility. Up to 65% of youth receiving MD treatment failed to respond, partly due to MD’s heterogeneity: an MD diagnosis reflected more than 1,400 possible symptom combinations, highlighting the need for treatments matched to personal need.

Treatment accessibility issues were similarly severe. Before the pandemic, less than 50% of youth with MD accessed any treatment at all; newfound financial strain further precluded families’ capacity to afford care for their children. It was thus critical to identify effective, scalable strategies to buffer against youth MD in the context of COVID-19, along with strategies to match such interventions with youth most likely to benefit.

This project integrated machine learning approaches and large-scale SSI research to rapidly test potent, accessible strategies for reducing adolescent MD during COVID-19.