A system described herein may use automated techniques, such as machine learning techniques, to analyze device snapshots from a group of User Equipment (“UEs”), and determine trouble conditions that are experienced by the UEs. The system may identify markers of the trouble conditions based on the snapshots, and may use these markers to predict or identify trouble conditions at other UEs based on snapshots received from the other UEs. Further, once a trouble condition is predicted or identified at the other UEs, the trouble condition may be proactively addressed, without requiring an explicit request from the other UEs to address the trouble condition.