Health Monitoring

Part of the Anaya Care Handbook — the source of truth for how the product must behave. When the product needs to change, change this document first, then make the system match it.

What this covers

This page governs how the platform tracks a client's health: vital signs collected during care visits, the per-client thresholds that decide when a reading raises an alert, symptom and side-effect tracking, wound care (the Anaya Healing Ally feature, including AI photo analysis), the acknowledgment of DNR and advance-directive orders by care providers, and doctors' appointments with reminders and AI visit-history summaries. The story in one paragraph: care providers record vital signs and symptoms as part of their care tasks; the platform checks every reading against thresholds set for that individual client and turns the entries into charts, trend lines, alerts, and AI-assisted period reports. Wounds get their own records and repeat assessments so everyone can see whether a wound is improving or worsening. When a client has a DNR order or advance directive, every care provider serving that client must acknowledge it. Doctors' appointments are scheduled with travel time, remind the care team automatically, and capture the visit outcome — including new prescriptions — when completed.

Change-of-condition health observations are a sibling feature covered in Change of Conditions. When a client leaves care, their health record follows the Transition of Care (discharge) process.

Key terms

• Vital signs — health readings recorded during care visits: weight, blood pressure, heart rate, oxygen saturation, blood sugar, body temperature, and respiratory rate.
• Abnormal reading — a vital sign outside the healthy range, flagged with a severity of low, high, or critical.
• Per-client vital threshold — the high/low limits for one vital on one client, set by a care manager in line with the client's healthcare-provider direction (physician orders), rather than a single range applied to everyone.
• Two-tier alert severity — the two alert levels a care manager can configure for each vital: an observation alert (note and watch) and a report-immediately alert (notify the care team without delay).
• Symptom log — a care provider's record of a symptom or medication side effect observed during a shift, captured per shift and kept as a chronological history.
• Symptom pattern — a recurring or escalating symptom the platform detects across a client's symptom history, which can raise an escalation alert at a defined threshold.
• Symptom report — a summary of a client's logged symptoms prepared for a healthcare provider.
• Period health report — an AI-assisted summary of a client's health over a chosen date range, optionally compared against a previous period; drafted, edited by a care manager, then finalized.
• Wound record — the file for one wound on one client: where it is on the body, its status, and its overall direction (trajectory).
• Wound assessment — one dated examination of a wound: measurements, tissue observations, pain level, photos, and notes.
• Wound trajectory — whether a wound is improving, stable, worsening, or not yet determinable, based on comparing recent assessments.
• AI wound analysis — an optional, on-demand analysis of an assessment's photo that produces a detailed description, estimated measurements, recommendations, warning signs, and a comparison with the previous assessment.
• DNR / advance-directive acknowledgment — a care provider's confirmation that they have reviewed a client's DNR order or advance directive.
• Stale acknowledgment — an acknowledgment made before the newest directive document was uploaded; the care provider must acknowledge again.
• Doctor's appointment — a scheduled visit between a client and a medical professional, with travel time, reminders, and an outcome captured on completion.
• Visit-history summary — an AI-written overview of a client's completed appointments, downloadable as a PDF.

How it works

Vital signs

Care providers do not enter vitals in a separate place — readings are captured as answers inside vitals care tasks (blood-pressure check, weight check, and so on) during their shifts (see Care Plans & Tasks). The platform reads those submitted entries to build everything else: the latest reading per vital, trend lines over time, and a list of abnormal readings. Every reading is checked against the thresholds and flagged when out of range, with a severity of low, high, or critical. A trend is called "increasing" or "decreasing" when the second half of the period differs from the first half by more than five percent; otherwise it is "stable".

Per-client thresholds. The high/low limits for each vital are meant to be set for the individual client, following the direction of that client's healthcare provider (physician orders) — they are never fixed platform-wide values. On top of those limits, a care manager can configure a two-tier alert severity for each metric: an observation alert (note and keep watching) and a report-immediately alert (notify the care team without delay). Both tiers are configured in line with the client's healthcare-provider direction.

Period health reports

A care manager can generate a report over a date range — either a single period or a comparison of two periods. The platform gathers the vitals, behavioral, and care-delivery data, asks the AI for a written summary and recommendations, and saves the result as a draft. The care manager can edit the draft (their edits sit on top of the AI's text), then finalize it. A finalized report is locked forever. Either version can be exported as a PDF, with the manager's edits taking precedence over the AI text.

Symptom & side-effect tracking

Care providers log symptoms and medication side effects during each shift, the same way they record vitals — as part of their care tasks. Each entry joins the client's symptom history, a chronological log anyone with health access can read. The platform watches that history for symptom patterns — symptoms that recur or escalate — and raises an escalation alert when a pattern crosses a defined threshold. Side effects are watched especially closely after a medication change. On request, the platform can produce a symptom report for a healthcare provider, and when a symptom is significant the client's representatives are notified.

Wound care (Anaya Healing Ally)

Each wound gets its own record, tied to a body location, with a status — Active, Healing, Healed, Chronic, or Referred — and a trajectory: Improving, Stable, Worsening, or Indeterminate. Closing a wound marks it Healed and records the date.

Care is documented through repeat assessments: measurements (length, width, depth), tissue observations, drainage, surrounding skin, pain level, treatment applied, and photos. After every new assessment, the platform compares the two most recent ones to recompute the trajectory: a wound area growing more than 10% means worsening, shrinking more than 10% means improving, a notable rise in healthy tissue means improving, and a notable rise in dead tissue always means worsening. A worsening wound alerts the agency's management team immediately, and every new assessment notifies management.

AI photo analysis is an on-demand product feature: once an assessment has at least one photo, anyone with wound-care access can request an analysis. The AI reviews the photo alongside the wound's recent history and the previous photo, then returns a detailed description, estimated measurements, tissue classification, care recommendations, warning signs of worsening, a comparison with the previous assessment, and a confidence level. The results are advisory — they sit on the assessment for clinicians to read and never change the wound's status or trajectory on their own. If the analysis fails, the assessment keeps all manually entered data and the analysis can simply be requested again.

DNR / advance-directive acknowledgment

When a client is flagged as having a DNR order or an advance directive, the platform shows a prominent, non-dismissible alert at the top of both the client profile and the care plan, and every care provider serving that client must acknowledge that they have reviewed it before their first shift with that client (see Scheduling & Shifts). The acknowledgment is per care provider, per client. If a newer directive document is uploaded afterward, the existing acknowledgment becomes stale and the care provider must acknowledge again. Acknowledging requires that a directive document is actually on file. Admins and care managers can review who has acknowledged for each client.

Doctors' appointments

An appointment links a client and a doctor (a medical professional user) with a start time in the client's timezone, travel time, transportation details, location, purpose, and optional pre-visit documents. The title is suggested automatically from the purpose, with a simple "Dr. Name – date" fallback. Recurring appointments (daily, weekly, monthly, yearly, with an end date or count) are created as individual occurrences that share a series identifier; the series is generated up to its end date or at most one year ahead, and changes apply to one occurrence at a time.

Travel time blocks the calendar both before and after the appointment. Before saving, the platform checks that blocked window against the client's shifts and care-plan tasks; overlapping task occurrences can be excused with the reason "doctor appointment" so care providers are not penalized for skipping them (see Scheduling & Shifts).

An appointment moves through four statuses:

Rescheduled appointments still count as upcoming and still receive reminders. Completing an appointment captures the visit outcome — diagnosis, prescriptions, notes, lab results, follow-up instructions, referrals, the next appointment date, and post-visit documents — and any newly prescribed medications are added to the client's medication list automatically (see Clients).

Reminders: when creating an appointment, the scheduler picks alert lead times (from 24 hours down to 5 minutes before). At each chosen time, the platform notifies the client's care team — that client's assigned care managers and assigned care providers — by in-app and push notification. Each person receives each reminder at most once. Representatives are not reminded: appointment reminders are a staff-coordination signal, and family notification is a separate decision (CL-34).

Visit-history summary: on request, the AI reads all of the client's completed appointments and writes a structured overview a doctor can review before the next visit. The newest summary is kept on the client and can be downloaded as a PDF. At least one completed appointment is required.

Rules

• HEALTH-1 — Vital-sign history must be derived from submitted care-task entries only; there is no separate vitals log, and unsubmitted or draft task entries never count.
• HEALTH-2 — The platform tracks exactly seven vital types: weight, blood pressure, heart rate, oxygen saturation, blood sugar, body temperature, and respiratory rate.
• HEALTH-3 — Every vital reading must be checked against the platform's healthy ranges and flagged as normal or abnormal, with abnormal readings carrying a severity of low, high, or critical.
• HEALTH-4 — Abnormality checks must respect the unit each reading was recorded in (for example pounds vs. kilograms, or the two blood-sugar scales).
• HEALTH-5 — Abnormal-reading lists, averages, and trends must cover all seven vital types.
• HEALTH-6 — A period health report can cover a single date range or compare two ranges, and is always saved as a draft first.
• HEALTH-7 — If the AI summary fails, the draft report must still be created; the AI text can be added later.
• HEALTH-8 — Care-manager edits to a report sit alongside the AI's original text and take precedence in the exported PDF.
• HEALTH-9 — A finalized report can never be edited or returned to draft.
• HEALTH-10 — Every wound is its own record on a single client, with a body location, a status (Active, Healing, Healed, Chronic, or Referred), and a trajectory.
• HEALTH-11 — Closing a wound must set its status to Healed and record the healed date.
• HEALTH-12 — After each new assessment, the wound's trajectory must be recomputed from the two most recent assessments: area growth above 10% means worsening, shrinkage above 10% means improving, a 10-point rise in healthy tissue means improving, and a 10-point rise in dead tissue always means worsening.
• HEALTH-13 — When a wound's trajectory turns to Worsening, the agency's management team must be alerted immediately.
• HEALTH-14 — Every new wound assessment must notify the agency's management team.
• HEALTH-15 — AI wound analysis can only be requested when the assessment has at least one photo, and never while an analysis for that assessment is already queued or running.
• HEALTH-16 — A failed AI analysis must never affect the manually recorded assessment, and the analysis can always be requested again.
• HEALTH-17 — AI analysis results are advisory: they must never change a wound's status or trajectory on their own.
• HEALTH-18 — When an AI analysis finishes, the person who requested it is notified.
• HEALTH-19 — Wound records and assessments are visible only within the client's own agency.
• HEALTH-20 — DNR / advance-directive acknowledgment is required exactly when the client is flagged as having a DNR order or an advance directive. (🚧 Spec only — the acknowledgment flow is not usable end-to-end yet; see Known gaps)
• HEALTH-21 — Each care provider has at most one current acknowledgment per client; uploading a newer directive document makes it stale, and the care provider must acknowledge again. (🚧 Spec only)
• HEALTH-22 — A care provider can only acknowledge when a directive document is actually on file for the client. (🚧 Spec only)
• HEALTH-23 — Every appointment must have a client, a doctor, a start time, a timezone, and a travel time; the title is suggested from the purpose with a "Dr. Name – date" fallback.
• HEALTH-24 — Recurring appointments are created as individual occurrences sharing one series, generated up to the series end date or at most one year ahead; edits, completions, and cancellations apply per occurrence.
• HEALTH-25 — Travel time blocks the calendar symmetrically before and after the appointment when checking for conflicts.
• HEALTH-26 — When an appointment is created, overlapping care-plan task occurrences within the blocked window can be excused with the reason "doctor appointment".
• HEALTH-27 — An appointment that is completed or cancelled can never change status again; only scheduled or rescheduled appointments can be completed, cancelled, or rescheduled.
• HEALTH-28 — Rescheduled appointments still count as upcoming and still receive reminders.
• HEALTH-29 — Completing an appointment captures the visit outcome, and newly prescribed medications are added to the client's medication list automatically.
• HEALTH-30 — Appointment reminders fire at each chosen lead time and notify the client's care team — that client's assigned care managers and assigned care providers (CL-34), not agency-wide management users and not representatives — with each person receiving each reminder at most once.
• HEALTH-31 — A visit-history summary can only be generated for a client with at least one completed appointment, and the newest summary must be downloadable as a PDF.
• HEALTH-32 — Appointments, like all health data on this page, must only be visible and editable within the client's own agency, by people granted the matching permission.
• HEALTH-33 — Every view of a client's health metrics or wound records is recorded in the access audit trail.
• HEALTH-34 — Vital-sign thresholds must be configurable per client in line with that client's healthcare-provider direction (physician orders), and never applied as fixed platform-wide values. (🚧 Spec only)
• HEALTH-35 — Each vital metric must support a two-tier alert severity — an observation alert and a report-immediately alert — that a care manager configures in line with the client's healthcare-provider direction. (🚧 Spec only)
• HEALTH-36 — Care providers must be able to log symptoms and medication side effects during each shift, and every entry must join the client's chronological symptom history. (🚧 Spec only)
• HEALTH-37 — The platform must detect recurring or escalating symptom patterns across a client's history and raise an escalation alert when a pattern crosses its defined threshold. (🚧 Spec only)
• HEALTH-38 — A symptom report must be producible for a healthcare provider, and a client's representatives must be notified when a significant symptom is logged. (🚧 Spec only)

Who can do what

Decisions needed

1. Should DNR acknowledgment gate clock-in? Today it is informational only. Options: block a care provider from clocking in to a shift until they have a current acknowledgment; show a blocking prompt at clock-in but allow override; or keep it as a passive requirement reviewed by managers.
2. Should AI wound findings drive alerts? The AI can report worsening indicators, but only manual measurements trigger worsening alerts today. Options: AI warning signs also alert management; AI findings stay advisory-only; or AI findings raise a "needs clinician review" flag without a full alert.
3. Should recurring appointment series be editable as a series? Today every change applies to a single occurrence. Options: support "edit/cancel this and all future occurrences"; keep per-occurrence editing only.
4. Vitals thresholds per client — resolved direction, not yet built. v4.2 settles that thresholds must be set per client from the client's healthcare-provider direction (physician orders), never as platform-wide values, with a two-tier alert severity per metric (HEALTH-34, HEALTH-35). Open question: who is allowed to enter and change those thresholds and the alert tiers — care manager only, or also a medical professional — and whether changes require a recorded physician-order source. Today the same ranges still apply to everyone (see Known gaps).
5. How significant is a "significant symptom"? Representatives must be notified when a significant symptom is logged (HEALTH-38). Open question: what defines significance — a fixed symptom list, a care-manager-set severity, or an AI judgement — and whether the representative notification is automatic or care-manager-confirmed.