Canonical governance page
Governance, consent, and AI transparency
This is the single governance page for every ICRBI program. Digital Twin Health 4.0, Tailor Care, Unity Impacts, the MSO, Care Console, and ScleroTrack all link here rather than publishing their own version, so these commitments are updated in one place and never drift.
A program of ICRBI — guided by the SPELL Impact framework
1. Consent and your choices
Nothing begins without your agreement. Before you join any ICRBI program, you will be told in plain language what information is collected, how it is used, and who may receive an alert about you. You can ask questions at any time, and you can change your mind.
- You choose to join. Participation is voluntary. Declining does not affect any other service you receive from ICRBI or from a clinical or community partner.
- You can pause. You may pause monitoring or outreach temporarily — for travel, illness, or any reason you do not have to explain.
- You can withdraw. You may withdraw at any time. We will tell you clearly what happens to your information when you do.
- You decide who else is involved. Caregiver and family access is granted only with your explicit permission, and you can change or remove it.
- Research is a separate decision. Consent to receive care support is never treated as consent to participate in research or a clinical trial. Those require a separate, explicit agreement.
2. What we collect, and why
We work to a minimum-necessary standard: we collect only what is needed for the care, evaluation, and operations you agreed to. If we cannot explain why a piece of information is needed, we should not be collecting it.
- Health readings such as blood pressure, weight, oxygen saturation, glucose, or activity — only where remote monitoring is part of your plan.
- What you tell us about pain, fatigue, mood, sleep, medications, and how you are managing day to day.
- Your goals in your own words — the outcome you personally want from the program.
- Practical circumstances such as transport, food access, language, caregiver availability, and internet access, so we can connect you to the right support.
- Clinical information from your care team, only where you have authorized that exchange.
Where clinical information is exchanged with a partner, we use HL7 FHIR, the standard application-programming interface for health information exchange, so that data movement is auditable and not locked into one vendor. Information collected for program measurement is held in our internal measurement layer only. It is never placed into public web analytics tools, and it is never sold.
3. Human clinical oversight
ICRBI is a community implementation layer. We are not a licensed clinical provider. Clinical care, diagnosis, treatment, and billing remain with licensed clinical partners.
Readings and patterns are reviewed by a registered nurse or trained care navigator. Our technology helps prioritize and summarize — it does not diagnose. Licensed clinicians make all clinical decisions. A risk score is never presented to you or to a clinician as a diagnosis.
4. What happens when a signal appears
Every category of signal has a named owner, an expected response time, a documentation method, and an escalation path. No alert is allowed to sit unassigned.
| What we notice | Who responds first | Where it escalates |
|---|---|---|
| A device or data problem | Digital navigator or technical support | Care coordinator if monitoring is interrupted |
| An everyday self-management opportunity | Coaching prompt, education, or peer support | Health coach or community health worker |
| A practical barrier — transport, food, benefits, caregiving | Closed-loop referral to a community partner | Care coordinator if the need stays unresolved |
| A clinically meaningful trend | Nurse review and protocol-based triage | Licensed clinician, or emergency pathway when warranted |
| A pattern affecting a whole group | Program review and resource planning | Community advisory council and partner action plan |
Referrals are closed-loop. We do not consider a referral complete when it is sent. It stays open until a partner confirms you actually received the help, and we measure how long that took.
5. Algorithm transparency and model cards
Where ICRBI uses a predictive or prioritization model, we maintain an internal model card for it. This approach follows the direction set by the federal HTI-1 rule, which established transparency requirements for artificial intelligence and predictive algorithms in certified health IT so that clinical users can assess them for fairness, appropriateness, validity, effectiveness, and safety.
Each model card records:
- Intended use — what the model is for, and explicitly what it is not for.
- Data sources — what the model was built and validated on.
- Performance — how well it does the job, measured honestly.
- Known limitations — where it is weaker, and which groups it has been tested least on.
- Bias testing — whether outputs differ systematically between groups.
- Update history — every version change and the reason for it.
- Responsible owner — a named person accountable for the model.
Model cards are reviewed at every model update and are available to clinical partners on request. Where a score is shown inside our applications, it links to its model card.
6. Equity by design
Equity is built into how the program is constructed, not added at reporting time.
- A low-tech path always exists. No internet and no smartphone is never a reason to be excluded. A community health worker can meet you in person, and you can reach us by phone.
- Language is planned, not improvised. Materials and support are offered in English, Arabic, French, and Spanish, and we measure how well language access actually performs.
- Caregivers are accommodated through explicit access rules rather than informal workarounds.
- Every measure is stratified. Our measurement records language, geography, disability status, insurance status, and digital-access level on every event, so no report can be produced without the ability to show who is being reached and who is not.
- Accessibility. Our sites are built to WCAG AA contrast standards and are fully keyboard navigable.
7. Community governance
A Community and Participant Advisory Council has meaningful influence over the program — not a ceremonial role. The council reviews program goals, public-facing participant language, communications, harm prevention, and how aggregated community data may be used, at least quarterly.
Where a council recommendation changes the program, we record what changed and why. Where we do not act on a recommendation, we record that too, and explain it back to the council.
8. Questions, corrections, and complaints
If you want to know what information we hold about you, correct something that is wrong, pause or withdraw, change who has access, or raise a concern about how you were treated, contact us. You will get a response, and raising a concern will never affect the support you receive.
Standards this page follows
These commitments are aligned with recognized frameworks rather than invented internally.
WHO framework on integrated people-centred health services, HealthIT.gov introduction to HL7 FHIR, HealthIT.gov HTI-1 final rule, WHO global strategy on digital health 2020–2025.
Governance page version 1.0 • August 2026 • Issued under the SPELL Impact Manifest v1.0. Reviewed at least annually and at every material change to consent, data handling, or model behavior.