Developmental Trajectory Modelling
Reading the shape of change, not the snapshot.
Early research into modelling a child's developmental path over years — detecting when a trajectory bends, and when a bend is worth a clinician's attention.
Unit of analysis
The trajectory, not the visit
Horizon
Multi-year
Open question
What counts as a meaningful bend
Stage
Early research
01Why snapshots under-serve development
In much of medicine a point-in-time measurement is diagnostically useful on its own. In neurodevelopment it is far weaker. What matters is the direction and rate of change, and whether that change responds to what is being done for the child.
A trajectory only exists if the same things are measured the same way over a long period and held in one continuous record. That is why this work sits downstream of the calibration project.
02The hard part is the threshold
Detecting that a curve has changed shape is tractable. Deciding which changes deserve to interrupt a clinician is the actual research question, and it is as much a clinical question as a statistical one.
An alert that fires too readily is worse than no alert, because it trains people to ignore it. We would rather under-alert and say so than manufacture sensitivity we cannot justify.
See what Deep Connection is really building.
Whichever side of this you're on, there's a way in.