All use cases

Engage / Lifecycle

Missed-dose recovery rules engine

A skipped weekly shot is not always the same problem — one missed dose needs reassurance, several in a row needs a clinician, and the agent knows which is which.

Close the loopGet aheadBehavioral triggerTime-based triggerPushChatSMSGLP-1 & PeptidesHealth & Telehealth

What it does

When a patient misses a GLP-1 dose, the right answer depends entirely on how much they missed and for how long. A single late dose inside the window is a straightforward take-or-skip call; but two or more consecutive misses change the picture, because restarting at the old dose after a gap can bring the nausea and GI side-effects roaring back, and the label points toward reconsidering the titration. Most programs treat every miss the same and either under-react or hand it all to a clinician. The agent runs the missed-dose logic as a branching rules engine: it reassures and guides the recoverable case from the label, and it recognizes the consecutive-miss pattern that needs a clinician to weigh re-titration — never deciding the dose itself.

How it works

  1. 1
    Trigger. a behavioral signal that an expected weekly injection was not logged, timed against the dosing window and the count of recent consecutive misses.
  2. 2
    Decision. the agent evaluates the missed-dose rules in order — inside the window, guide a safe resume per the label; past it, skip and resume on the usual day; multiple consecutive misses, treat as a re-titration question and route it. It reads memory for the patient's tolerance history so the escalation carries the GI-rebound context a clinician needs.
  3. 3
    Action. for the recoverable branch, a reassuring, label-grounded message on push, chat, or SMS that closes the loop; for the consecutive-miss branch, a patient-facing "let's get your clinician to confirm the safest way to restart" plus a structured handoff to the care team.
  4. 4
    Follow-through. the outcome and any restart plan write back to memory; the clinician's decision is relayed to the patient and the loop is closed; if the patient resumes on their own inside the window, the pending escalation is cancelled. Repeated entries into the engine feed the adherence picture the program tracks against a holdout.
Example
AgentI noticed this week's dose hasn't been logged, and it looks like last week's was missed too. After two in a row, the safest restart can depend on how your body handled the medication before — I don't want you to guess at that.
CustomerYeah, I ran out and missed a couple. Can I just pick up where I left off?
AgentBecause it's been more than one dose, your clinician should confirm whether to resume at your last dose or step back to ease back in — restarting too high after a gap can bring the side-effects back. I've sent them your dose history; I'll bring their answer straight back to you here.

Configuration

How the agent is wired for this use case.

Triggera dose_missed behavioral signal from the app backend (no injection log against an expected weekly dose), evaluated with the dosing window and a running consecutive-miss count.
Tools & actions
  • App backend · provides the expected dose schedule, missed-dose events, and the consecutive-miss count.
  • Patient memory store · reads tolerance and titration history for the escalation context; writes the outcome and any restart plan back.
  • Rules / knowledge base · encodes the label's branching missed-dose logic (resume-in-window, skip, consecutive-miss re-titration question).
  • Messaging channel · delivers the reassurance branch or the escalation message on push, chat, or SMS.
  • Clinician escalation queue · receives the consecutive-miss case with dose history and GI-rebound context for a re-titration decision, and returns the decision to relay.
Autonomythe recoverable branch runs unattended after a quality judge clears the label-grounded message; the consecutive-miss branch always routes to a clinician (detect, then escalate). The engine never selects, changes, or recommends a specific dose — re-titration is a clinician decision, and patient-facing text stays educational.
Channelspush · chat · sms
Escalationtwo or more consecutive missed doses, or any symptom or distress in the reply, hands off to a clinician with the patient's dose and tolerance history attached.

What you need

The inputs this use case runs on. Your channels stay yours; the agent supplies the judgment.

Signals

missed-dose events, the dosing-window definition, the running consecutive-miss count, injection logs.

Data

dose schedule and medication on file, the label's missed-dose and re-titration rules, the patient's tolerance and titration history, consent state.

Guardrails

clinical detect-to-escalate on every consecutive-miss case — no dose selection or re-titration advice from the agent; patient-facing text stays educational and label-grounded; judge gating on the unattended branch; frequency caps; cancellation of a pending escalation if the patient safely resumes; holdout measurement of the adherence effect.

Metrics it moves

  • adherenceup, as the recoverable miss is resolved cleanly and the risky gap gets the clinician it needs instead of an unsafe self-restart.
  • churndown, because a string of missed doses is caught and routed before it becomes a quiet drop-off.
  • ticket-deflectionup, the common "I missed a few doses, what do I do?" question is triaged automatically, with only the cases that need a clinician reaching one.

Related use cases

Engage / Lifecycle

Weekly-injection reminder with missed-dose next-step

the same-day reminder and single next-step this engine deepens on repeated misses

Prevent / Care

Dose-hold coaching on a side-effect spike

the sibling that surfaces a hold to the clinician on a tolerance signal

Prevent / Care

Clinical red-flag detection and safe escalation

the safety rail the consecutive-miss handoff relies on

See it on your own customer journey

Bring one drop-off, one churn cliff, or one silent segment. We will show you what a proactive agent with memory and judgment does with it.

Book a demo