[problem: heterogeneity, no prognostic value, responder non-resonder, we're working with a classificationsystem that doesn't work.]
[biology: network models are everywhere, explain fundamentals of everything, in psychiatry now too, can we use it?]
Research question: can we use the network of depressive symptoms as a meaningful phenotype? 1. Can we
reliably estimate network strength independent of actual symptom levels. 2.
Does network strength differentiate between:
depressed versus matched never-depressed adolescents
For responders, acutely ill state versus recovered state
At baseline, future responders versus future non-responders
3. Is network
strength associated with other (non-depression) outcomes?