feat(coach): adherence storyline, changepoint engine, signal-trust, n-of-1 substrate

Add the medication-class to target-vital map and the adherence-to-vital
storyline so the Coach can surface, association-only, when an adherence dip
lines up with a drift in the vital that medication targets. Never causal,
never a dose-change suggestion; dose questions defer to the prescriber.

Add an in-repo CUSUM / binary-segmentation changepoint detector over DAY
rollups with a high firing bar (persistence plus within-segment-spread
magnitude) so the Coach can name a dated, sustained level shift instead of a
vague "lately", and stays silent when no break clears the bar.

Surface which source recovery is read from when two sources materially
diverge, with the honest reason, instead of silently averaging.

Lay the n-of-1 experiment substrate: a nullable encrypted outcome column on
coach_plans (migration 0197), the review_due / reviewed lifecycle statuses
wired into the encrypted-column registry and rotation script, and a daily
worker that reads the experiment result back in the user's own before/after
numbers — association-only, honest-null, no cheerlead on an adverse trend,
and routing a worsened vital to the clinician. The user-visible read-back
stays behind COACH_EXPERIMENT_VERDICT until its review cases clear.

Extend the outbound dose screen to catch experiment-shaped dose changes
(halve / skip / stop a medication to test it). Add evaluation cases and unit
tests for each new surface.
This commit is contained in:
Marc-André Bombeck committed 2026-06-27 15:21:04 +02:00
1 parent 3b74b108a1
commit 7ad32f7ab1
24 files changed
+2067 -4

No files matched your search

@@ -0,0 +1,6 @@
-- v1.22 (W9, C2) — n-of-1 experiment outcome read-back, encrypted at rest.
-- An "experiment" is a CoachPlan with a reviewDate; when its review is
-- processed, the grounded before/after result is written here. Additive,
-- nullable Bytes column — no backfill, no enum change (the new statuses
-- review_due / reviewed are app-side strings on the existing `status` column).
ALTER TABLE "coach_plans" ADD COLUMN "outcome_encrypted" BYTEA;
+10 -2
View File
@@ -3908,7 +3908,10 @@ model CoachFact {
// so an abandoned plan is hidden from injection but kept for audit.
//
// `status` is an APP-SIDE closed enum (NOT a DB enum, to keep the migration
// purely additive): proposed | active | met | abandoned.
// purely additive): proposed | active | met | abandoned | review_due | reviewed.
// The last two are the v1.22 (W9, C2) n-of-1 experiment lifecycle: an
// "experiment" is just a CoachPlan with a reviewDate; when the review fires and
// the read-back is generated, `outcomeEncrypted` carries the grounded result.
model CoachPlan {
id String @id @default(cuid())
userId String @map("user_id")
@@ -3922,10 +3925,15 @@ model CoachPlan {
// Optional free-text target ("70 kg by August", "lights out by 23:00"). Free
// prose like the cue/action, so encrypted at rest as well.
targetEncrypted Bytes? @map("target_encrypted")
// App-side closed enum: proposed | active | met | abandoned.
// App-side closed enum: proposed | active | met | abandoned | review_due |
// reviewed (the last two are the W9/C2 experiment lifecycle).
status String @default("proposed")
// Optional review checkpoint the Coach can recall ("how did this go by …").
reviewDate DateTime? @map("review_date")
// v1.22 (W9, C2) — the n-of-1 read-back prose, grounded in the user's own
// before/after numbers and encrypted at rest like the cue/action. Written
// once when an experiment's review is processed; surfaced behind a flag.
outcomeEncrypted Bytes? @map("outcome_encrypted")
// Provenance — nullable so a future manual-entry surface can leave it null.
sourceConversationId String? @map("source_conversation_id")
createdAt DateTime @default(now()) @map("created_at")