> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kaireonai.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Healthcare & Life Sciences
> Consent-first reminders and program nudges — appointment prompts, refill reminders, and wellness invitations — delivered respectfully and never clinically.
KaireonAI orchestrates **administrative and engagement communications** — reminders, invitations, and program nudges — not clinical or medical advice. Every example below is a scheduling or enrollment prompt. Decisions about care belong to clinicians and patients; the platform's role is to send the right respectful reminder, to the right person, on a channel they've consented to.
## The problem in this industry
Healthcare organizations reach members constantly — appointment reminders, refill prompts, wellness-program invitations, portal-activation nudges — and the volume can easily tip from helpful into intrusive. A member who gets four reminder texts in a week for something they've already handled stops reading all of them, including the one that mattered. And because health communication is sensitive, the cost of getting consent, timing, or frequency wrong is not just a missed engagement; it's a trust and compliance problem.
The two hard constraints are consent and restraint. A member must have opted into the channel before you use it, opt-outs must be honored immediately and permanently, and quiet hours must be respected. Within those constraints, the goal is coordination: the reminder team, the wellness team, and the portal team should not each independently message the same member, and no single member should be over-contacted just because several well-meaning programs all qualified them at once.
KaireonAI is consent-first by design. Every decision checks recorded [consent](/governance-security/consent-management) before a channel is used, honors do-not-contact permanently, respects time-of-day windows, and enforces one shared contact budget across all programs — so a member receives coordinated, welcome nudges rather than a barrage.
## What you build in KaireonAI
You express your reminder and program communications as a small set of platform building blocks, with consent and restraint wired into every layer. Here is how a typical member-engagement setup maps on:
| Platform concept | How you use it in healthcare engagement |
| ------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ |
| **Offers** (engagement prompts) | Annual-checkup reminder, flu-shot clinic invitation, wellness-program enrollment, telehealth-availability notice, prescription-refill reminder, patient-portal activation, care-plan check-in prompt, screening-eligibility notice |
| **Channels** | Patient portal, secure email, SMS (consented only), app push |
| **Decisioning gates** (Eligibility → Fit → Match) | *Eligibility*: `attribute_condition` on `consent_on_file == true` and `program_eligible == true`; suppress a reminder for something already completed (`offer_attribute` / `recency_check`). *Fit*: `segment_required` for program-specific invitations. *Match*: gentle prioritization of the most timely reminder |
| **Contact policies** | `do_not_contact` for opt-outs (honored permanently); `time_window` to enforce quiet hours; `frequency_cap` set conservatively; `customer_total_cap` so all programs share one respectful weekly budget; `cooldown` so a reminder isn't repeated too soon |
| **Scoring approach** | Favor transparency: a [`scorecard`](/ai-ml/algorithms/scorecard) or [`logistic_regression`](/ai-ml/algorithms/logistic-regression) makes it clear *why* a member received a given reminder, which matters for oversight. Reserve faster-learning models for low-sensitivity prompts like a general wellness-program invitation |
The [consent stage](/governance-security/consent-management#enforcement-at-decision-time) runs on every recommendation and suppresses candidates for any channel whose consent was revoked — an independent backstop beyond the eligibility gate. [DSAR and data-portability](/governance-security/dsar-portability) tooling supports member data-access and erasure requests.
## A worked example
**Member M-2048** has consented to SMS and portal messaging (not email), is eligible for the annual-checkup reminder and a flu-shot clinic invitation, activated their portal last month, and received a refill reminder yesterday. It's currently outside quiet hours.
Checkup reminder, flu-shot invite, wellness enrollment, telehealth notice, refill reminder, portal activation, care-plan check-in, screening notice.
Portal activation drops (already activated last month — `recency_check`). The refill reminder drops (sent yesterday). Screening notice drops (not yet eligible). Five survive.
The consent stage removes any candidate that would need email (no email consent). A conservative `frequency_cap` and the shared `customer_total_cap` hold back lower-priority prompts so the member isn't over-messaged this week.
A transparent scorecard ranks the survivors. The **flu-shot clinic invitation** ranks first — timely (in season), eligible, and not recently sent — over the more evergreen wellness-enrollment prompt.
The member gets a single, consented, well-timed nudge — during allowed hours — rather than several overlapping reminders. The [decision trace](/api-reference/decision-traces) records the consent check and every suppression for audit.
**Consent is the deciding factor, every time.** Had M-2048 revoked SMS consent, the same decision would either route to the portal instead or send nothing — the platform never falls back to a channel the member didn't agree to. And a member who opted out entirely is removed by `do_not_contact` before scoring even runs, permanently, with no exception path.
## Measuring success
* **[Business Dashboard](/operations-reporting/dashboards#business-dashboard)** — engagement rate per prompt type (e.g. how many invitations lead to a scheduled action).
* **Suppression rate** — a healthy, conservative program suppresses a meaningful share of would-be sends; watch this closely, since under-suppression is the risk here, not under-sending.
* **Consent and opt-out integrity** — confirm via decision traces that revoked-consent channels are consistently excluded.
* **Uplift and holdout** — an [experiment](/ai-ml/uplift-modeling) with a no-reminder control group measures whether nudges genuinely increase completed actions (scheduled appointments, refills picked up) versus actions that would have happened anyway.
## Where the agentic layer helps
[Decision Sentinel](/ai-ml/sentinel) is especially valuable in a consent-heavy setting: a mis-scoped policy or a broken quiet-hours window can silently stop reminders — a member who *should* have gotten a timely prompt gets nothing, with no error to signal it. Sentinel watches the decision stream for empty-decision surges and suppression spikes and alerts before a program goes dark.
[Decisioning Autopilot](/ai-ml/autopilot) can propose improvements to non-clinical engagement models — for example promoting a challenger that lifts wellness-invitation response — always as reviewable, four-eyes-gated proposals so a human owner signs off before anything changes. The [governed AI assistant](/ai-ml/ai-assistant) lets program owners adjust reminder cadence in plain language, routed through approvals, keeping every change auditable.
## Try it
A staged, stop-when-done nudge sequence — the same pattern fits portal activation and program enrollment.
Record consent, honor opt-outs permanently, and enforce channel preferences at decision time.
Quiet hours, conservative frequency caps, and a shared per-member contact budget.
Register and build a consent-first reminder flow end to end.