Behavior trends
It reads incident exports and ABC logs, finds the recurring trigger, and checks it against the behavior support plan.
GER export · 30 days
10 events · 3 individuals
falls, med errors, behavior
remy.
finding the pattern
8× sensory trigger
Loud appliances precede the behavior, eight times out of ten.
Matches the BSP
Consistent with the plan's hypothesized function: escape.
Fidelity gap
Headphones weren't offered first in the latest event.
Works with Therap GER exports, or CareNotes ABC logs (I/DD)
01
The GER export Therap already produces, or ABC events logged in CareNotes during visits. Thirty days is enough to start seeing shape.
02
Recurring antecedents across events - the blender, the schedule change, the loud day room - counted and named, with each underlying event linked.
03
Does the pattern support the BSP's hypothesized function? Was the proactive strategy actually used before the last event? Your BCBA gets analysis, not anecdotes.
Across 10 logged events
8× loud appliance noise → behavior
matches the BSP: escape from sensory input
→ headphones first, before the blender
'Loud/aversive sensory input, 8 of 10 events' beats 'he has bad days sometimes.'
Patterns are checked against the plan's hypothesized function - support or revisit, stated plainly.
Flags when the plan's proactive strategy wasn't used before an event. That's coaching gold.
Every claim links to the DSP-written events behind it. Nothing is asserted without its source.
8 of 10
events shared one antecedent in the sample export - invisible one event at a time.
30 days
of events is enough for the first meaningful pattern read.
100%
of findings link to the underlying logged events.
01
The standing review starts from patterns, not a stack of forms.
02
Evidence for or against the hypothesized function, continuously.
03
'Offer headphones before the blender' - to the right staff, about the right moment.
04
A rising pattern gets seen this month, not at the annual review.
| By hand | Generic AI | remy. | |
|---|---|---|---|
| Pattern finding | Binder + memory | Summarizes one report | Counts across every event |
| BSP connection | Rarely revisited | Doesn't know the BSP | Checked against the stated function |
| Evidence | Anecdote | Unsourced | Every claim links to its events |
| Timing | Quarterly, maybe | - | Every export |
No. It reports whether logged events support the plan's stated hypothesis. Interpreting that - and changing any plan - belongs to your BCBA and the team.
No. State reporting stays in Therap, where states expect it. remy reads the export to find patterns - it writes nothing back.
It will analyze whatever you give it, and it says when the data is too thin to call a pattern - 'insufficient data' is an answer it's allowed to give.
The trends analysis is built for I/DD workflows - ABC data, BSPs, GERs. Home-care agencies get the same engine pointed at visit patterns via the briefing.
The trial workspace opens with demo data already in it. 14 days free, no card.