Evi's files showed thirty-one active households in her coordination district. Most were in the green range: sleep-signal indices stable, no flagged patterns, no intervention recommended. She checked them on Tuesday mornings, the way a doctor checks bloodwork, looking for the numbers that stood out.
The Halvorsen household had been flagged for the third consecutive month.
She pulled the full index. Two adults, one child, age eight. The adults' sleep-signal profiles showed what the classification system called departure-pattern ideation: a cluster of dream-content markers associated with leaving, spatial movement, boundary crossing, reduction of possessions, that taken individually meant nothing, but in sustained concentration over eleven weeks indicated active, unresolved cognitive processing of exit scenarios. The child's sleep quality index had dropped from a baseline of 82 to 51. The child was not sleeping badly, exactly. The child was sleeping differently, in a way the algorithm associated with ambient household conflict.
No incident had been reported. No request for intervention had been filed. The Halvorsens, as far as any official record showed, were a functioning household.
The algorithm had flagged them because the dreams said something was not right.
Evi's role was coordinator, not counselor. She was not authorized to initiate contact without a request. The flag was a recommendation, not a directive. She had the right to set it aside.
She looked at the child's index again. Fifty-one. The baseline had been 82 four months ago.
She opened the household contact file.