← Back

Guide · Published November 14, 2026 · Reviewed November 14, 2026 · Ingenious AI team · Not legal advice

← All guides

ADMT Notices for Healthcare Services Decisions: Coverage, Access, and Explanations

The short answer: Healthcare services sit on both big US lists in the kit's data. Colorado's SB 26-189 covers consequential decisions about health-care services: pre-use notice at the point of interaction, 30-day adverse-outcome explanation, data correction, meaningful human review, and 3-year records, all from January 1, 2027. California's significant-decision list includes healthcare services: risk assessment before use, pre-use notice, opt-out where given, and access answers explaining the logic, outcome, and purpose. This guide covers the kit's duties only — it is not a guide to health law generally, and the kit is a templates and records tool, not legal or medical advice.

Check your duties and dates — free Build your first notice — free

What is a healthcare services decision here? A decision that results in the provision or denial of healthcare services (California's wording) or a consequential decision about health-care services (Colorado's list) — eligibility, access, scheduling into care pathways, or service authorisation steps where an automated tool materially influences the outcome. Clinical judgment exercised by a clinician is a different matter from an access decision supported by a tool; test your workflow in the free checker.

Where the tool actually sits in healthcare workflows

Healthcare's automated decisions cluster at the edges of care rather than inside the consultation: eligibility and enrolment screening, prior-authorisation support, appointment and pathway allocation, and service-denial workflows. Those edges are exactly where the kit's lists point — provision or denial of services. A triage-support tool that materially influences who is seen, when, or whether a service is authorised is closer to the line than a purely administrative reminder system.

The factor question from the checker does real work here: if the tool is one minor input and a person genuinely decides, Colorado's duties do not attach to incidental use; if the tool ranks, narrows options, or sets terms, it has crossed into material influence. Document which side each system sits on, with the reasoning, in the inventory — see How to Tell If a Tool Is ADMT.

Colorado healthcare: the five duties at the service door

For a covered Colorado use, the sequence is the standard five in a healthcare key: pre-use notice at the point of interaction — the enrolment flow, the portal, the intake step — before the tool is used; 30-day explanation after an adverse outcome, in plain language a patient can act on; correction of the data the tool used (health-adjacent data errors are common and consequential); meaningful human review by a person with authority to change the outcome; 3-year records.

Plain language carries extra weight here. An explanation full of internal codes fails the purpose even if it meets the date: the person must be able to identify what to correct or what to ask review to reconsider. The drafting standard is worked through in ADMT Notices in Plain Language.

Worked example: Prairie Health Network

Example — fictional business, real duty structure. Prairie Health Network is a fictional regional provider using a tool to support service-eligibility decisions in Colorado and California. One quarter under the duties:

California healthcare: assessment, notice, opt-out, access

California's set applies in its usual order: risk assessment first (purpose, data involved, risks, safeguards — in healthcare the data element will be sensitive by nature, which raises the stakes on the safeguards element), pre-use notice describing the use and naming opt-out and access rights, two easy opt-out routes where the opt-out applies, and access answers covering logic, outcome, and purpose on request.

An opt-out in a services context means a human route to the same decision. For time-sensitive services that route needs to be real — staffed, reachable, and not slower in a way that punishes the person for using it. Design it as a care pathway, not a compliance corridor. The opt-out design rules are in California ADMT Opt-Out Requirements.

What this guide and the kit deliberately do not cover

The kit's checker covers the automated-decision duties in its data — Colorado, California, Connecticut, and the EU Article 50 / Annex III material — and nothing else. It does not cover clinical regulation, health-data law, or insurance law, and the kit never judges whether a decision was fair or clinically right. A healthcare operator will have other regimes in play; this guide's claim is only that these notice, explanation, and record duties are among them, on the dates stated.

Keeping that boundary clear is protective, not limiting: the record the kit keeps — notices, explanations, opt-outs, reviews, with dates — is the evidence base for these duties, and it neither claims nor needs to claim anything about the quality of care. Other questions go to other systems and other advisers.

Connecticut and the EU, briefly and accurately

Connecticut's profiling list includes health care services among its significant effects, so a covered controller's profiling duties — opt-out answered within 45 days, outcome questions, data-protection assessment — can attach to healthcare profiling outside employment. Connecticut's employment-specific AEDT rules, by contrast, expressly leave out health and safety decisions. The distinction is in the two Connecticut tracks, explained in Connecticut Profiling Opt-Out Rights.

In the EU, healthcare is not one of the Annex III areas the kit's checker data names (employment, education, creditworthiness, essential services) — this guide does not extend the list. Article 50's interaction and content duties, however, attach regardless of sector since August 2, 2026, so a patient-facing AI system carries those today. Precision about which list you are on is the whole discipline here.

  1. Locate the tools at the edges of care — eligibility, authorisation support, pathway allocation — and test each one's role honestly
  2. Colorado: run the five duties with explanations written for patients, not for internal readers
  3. California: complete the risk assessment with a safeguards element that matches the sensitivity of the data
  4. Give the opt-out a human route that is genuinely usable for time-sensitive services
  5. Keep the boundary clear in your own documentation: this record evidences the notice duties, nothing about clinical quality

Sources, as carried in the kit's own data: the duty summaries in app/rules.py and the checker data in app/checker_data.py, verified against primary or authoritative sources on October 28, 2026. Run the free duties checker for your jurisdiction, decision type, and the tool's role, and compare the kit's plans for the record, builder, and review queue. More from Ingenious AI is on the hub.

Build your first notice — free

Related guides

See all ADMT guides for the full set.

Frequently asked questions

Are healthcare services decisions covered by ADMT rules?

Yes, on both US lists in the kit's data: health-care services on Colorado's consequential-decision list, and healthcare services on California's significant-decision list.

What must a Colorado provider send after denying a service?

A plain-language explanation of the principal reasons and the tool's role within 30 days of the decision, with correction and human-review paths named.

Does a clinician's judgment count as automated decision-making?

The checker asks how big the tool's role is. If a person genuinely decides and the tool is one minor input, Colorado's duties do not attach to incidental use. If the tool materially influences the outcome, they do.

Is healthcare on the EU Annex III list in the kit's data?

The Annex III areas the kit's checker data names are employment and worker management, education access, creditworthiness evaluation, and essential public and private services. This guide does not extend that list.

Is the ADMT Notice Kit medical or legal advice?

No. It is a templates and records tool: it generates the notices and explanations, runs the review queue, and keeps the record. It never judges whether a decision was fair.

Build your first notice — free