Skip to content
LifeWatch AI

Trust

Where the system stops.

Every governance committee asks the same seven questions, and none are answered by a badge. This page answers them with mechanisms — and says plainly where we are narrower than the question invites.

The boundary

What it does. What it never does.

Stated once, in full, because everything else on this page depends on it holding.

What it handles

  • Asking what the program says to ask, in the patient's own context
  • Answering questions the program has been approved to answer
  • Booking, confirming, and reminding about appointments
  • Routing refill problems to the queue that owns them
  • Delivering education and checking that it landed
  • Recording adherence, symptoms, and circumstances against the record
  • Raising anything that meets the program's escalation criteria

What it never does

  • Diagnose, or suggest what a symptom means
  • Prescribe, change a dose, or advise stopping a medication
  • Make or influence a treatment decision
  • Give medical advice of any kind
  • Handle an emergency — patients are told to call 911
  • Decide its own escalation rules

The crisis pathway is live on every contact. A patient describing an emergency is told to call 911, and the call does not continue around it.

The proof

Nothing is raised that cannot be traced to something the patient said.

This is the whole of what a nurse receives: the triggering sentence marked, the change measured against that patient's own baseline, and the rule from the certified program you read before go-live.

Escalation · delivered to the care team

UrgentRuth Alvarez · 74 · MRN 4471902

Weight rising since discharge with new breathlessness at night

Delivered

Within a minute of the call ending

Why this was raised

Triggering statement

I have been waking up needing to sit forward. And my shoes are tight on me again.
Change from baseline
Four pounds above the weight recorded at discharge six days ago.
Sent to
Her own care manager, with the covering clinician copied

Escalated because it meets criteria written into this program by our clinical team and reviewed with your organization before go-live. LifeWatch applies the rule; it does not decide what the rule should be.

What the care team received

  • The transcript, with the triggering statement marked
  • The change against her own discharge baseline
  • Her conditions, current medications, and the discharge medication change
  • What the call covered, and what it deliberately did not answer
  • Why the program treated this as urgent

What LifeWatch did not do

  • No diagnosis was offered
  • No medication or dose was changed
  • No treatment advice was given
  • The patient was told a nurse would call, and nothing more
For the governance committee: escalation mechanics and the seven questions
Escalation mechanics

What creates one, what travels with it, where it goes.

01What creates one
Findings defined in the program itself, plus anything the patient raises outside what the call is scoped to handle. The criteria are readable content, not model behavior — you can open the program and point at them.
02What travels with it
The transcript, the triggering sentence, the patient's relevant context, and the change against their own baseline. The reviewing clinician never starts from an alert with no reason.
03Where it goes
Into the queue your organization chose for that program, ranked by severity — always the patient's own care team, never an outside clinical service.
04What you can check afterward
Every contact keeps its transcript and what was derived from it, and what happened returns to the patient's record.
Governance

The seven questions, including the two we answer narrowly.

Two of our answers stop short of what the question invites. We would rather you read that here than discover it in diligence.

  1. 01Is the system clinically validated?

    Every program is authored and certified by our clinical team before it runs. Escalation behavior is measured against a fixed rubric before any change ships.

    Where we stop short. We do not claim independent clinical validation or published outcome evidence. No such study exists yet, and we will not describe internal evaluation as though it were one.

  2. 02Can you explain why it did what it did?

    Every escalation carries the exact sentence that caused it, the change against that patient's own baseline, the program rule that applied, and the full transcript.

  3. 03Is there governance and an audit trail?

    Access to a patient record is logged with actor, action, time, and outcome. The audit log is append-only, enforced at the database level — entries cannot be edited or deleted, including by our own application. Signed notes are made immutable the same way.

  4. 04How is bias monitored?

    Reach, completion, and escalation rates are tracked by cohort, so differences between groups surface in the operational reporting rather than staying buried. The evaluation set spans the ways patients actually describe symptoms — including through an interpreter or a family member.

    Where we stop short. This is monitoring and reporting, not a certified fairness audit. We would rather tell you what we measure than imply an assurance nobody has given us.

  5. 05Where is the human oversight?

    Our clinicians decide what every program covers and what must escalate, before any patient is contacted. Anything meeting those criteria goes to the patient's own care team, and the boundary above holds on every contact.

  6. 06How are changes managed?

    A changed program or rubric is re-evaluated and re-certified before release — the gate shown above. You are told what changed before it reaches your patients.

  7. 07What frameworks do you work to?

    HIPAA governs how the system is built and operated, and every customer relationship runs under a business associate agreement. We track the emerging health-AI governance frameworks and answer their questions in these terms — mechanisms rather than membership.

Bring your hardest question.

The conversations we do best are with the clinician who expects this to be a bad idea. Ask about the boundary, the escalation rule, or the thing you have seen go wrong before.

Talk to our team