Editorial standards.

Most clinical reference apps anchor their credibility on a named medical-doctor advisor. NurseMind doesn't. Instead, we anchor on three things, in this order:

1. License-correct foundation sources. Every clinical statement in the library is sourced from public-domain or openly-licensed primary material from accredited authors — federal regulators, nursing faculty, public health agencies, peer-reviewed open-access journals.
2. Cite-or-refuse enforcement. Citations on every clinical claim, validated server-side. Numerical claims without a citation are rejected by our pipeline before they reach you.
3. Tiered RN review. High-risk content is reviewed by licensed registered nurses pre-publication. The rest is sampled. Every entry has a feedback control.

We are not the originator of clinical knowledge. We are the shelf, the index, and the search engine. The decision to apply anything from this product in a clinical setting remains with the licensed clinician at the bedside.

The library is built only from sources whose license terms permit commercial use with attribution. Each entry cites the specific source and complies with that source's attribution requirements.

• openFDA / DailyMedPublic domain
  FDA-approved drug labeling — indications, dosing, contraindications, black-box warnings, drug-interaction data
• RxNormPublic domain (NLM)
  Standardized drug naming — generic names, brand names, ingredient mappings
• Open RNCC BY 4.0
  Wisconsin Technical College System nursing textbooks (Pharmacology, Med-Surg, Mental Health, Nursing Skills, etc.)
• OpenStax NursingCC BY 4.0
  Rice University open nursing textbooks (Pharmacology for Nurses, Maternal-Newborn, etc.). Library display only — AI corpus use pending OpenStax permission.
• VA PBM MonographsPublic domain
  U.S. Department of Veterans Affairs Pharmacy Benefits Management drug monographs
• LiverToxPublic domain
  NIH National Institute of Diabetes and Digestive and Kidney Diseases — drug-induced liver injury reference
• CDCPublic domain (U.S. federal work)
  Centers for Disease Control and Prevention — infection control guidance, vaccination schedules, treatment guidelines
• NIH treatment guidelinesPublic domain
  Disease-specific clinical practice guidelines from NIH-supported workgroups (e.g., COVID-19 Treatment Guidelines)
• MedlinePlus (NLM-authored)Public domain (NLM-authored sections only)
  National Library of Medicine consumer health pages — for definitions and patient-education context
• PubMed Central Open AccessCC BY / CC0 (subset)
  Peer-reviewed journal articles licensed CC BY or CC0, used for evidence updates beyond textbook references

Some references are well-known in nursing but their licenses don't permit the curator model — either because they prohibit commercial use, derivative works, or AI ingestion, or because they are proprietary works available only by paid subscription. We do not republish from any of these:

StatPearls— NC-ND license
AHFS PMI— proprietary
MedlinePlus drug pages— A.D.A.M. third-party content
Testing.com— proprietary
DrugBank (full database)— commercial license required
Davis's Drug Guide— proprietary publisher
Mosby's— proprietary publisher
Lexicomp— proprietary publisher
UpToDate— proprietary publisher
Nursing Central— proprietary publisher
ATI / Saunders / Lippincott / Kaplan / UWorld— proprietary review-book publishers
Live NCLEX items— NCSBN trademark + content protection

The omission is not a quality judgment. Many of these are excellent references — we just can't redistribute them under our model. If you study from them in other contexts, that's between you, them, and your school.

Every clinical assertion in the library is tied to a specific source. Every numerical claim in an Ask NurseMind response is tied to a specific source. The pipeline enforces this:

• Library entries are authored as a curated assembly of citation-tagged blocks. Each dose, range, threshold, contraindication, or monitoring parameter has a source attached at the data-model level — there is no text-only path that doesn't carry a citation reference.
• AI co-pilot responses are validated server-side via regex enforcement before being sent back to the device. Numerical claims (doses, ranges, thresholds) must include a square-bracket citation that resolves to a chunk in the retrieved context. Responses that don't comply are rejected and re-generated.
• If we can't cite it, the entry is not shipped or the AI returns a refusal explaining what it would have needed to answer.

When you ask a question, NurseMind retrieves the most relevant entries from the curated library using keyword and semantic matching. The AI is given only those retrieved entries plus a tightly-scoped system prompt that instructs it to:

• Answer using only the supplied retrieved context — never external pretraining knowledge — for any clinical claim.
• Include citations on every numerical claim, with the citation IDs that appear in the supplied context.
• Refuse to give a diagnosis, prescribe, direct treatment for a specific patient, or process patient-identifying information.
• Stay within nursing-reference scope (drugs, drips, labs, scenarios, communication, pathophysiology, NCLEX content) — questions outside that scope receive a polite redirect.

Inputs are passed through a server-side scrubber before reaching any third-party model. The scrubber detects and removes apparent patient-identifying data (names, MRNs, dates of birth, room numbers). Inputs that contain such data are rejected, and the user receives a refusal asking them to describe the situation in general terms.

AI responses can still be wrong. Models hallucinate, retrieval misses relevant evidence, and even cited sources can be misread. We don't pretend otherwise. Verify against the cited primary source before relying on anything in any context where accuracy matters.

We don't have a full-time clinical advisor. Instead, we engage a pool of paid licensed registered nurses for review in three tiers, scaled to risk:

NurseMind is not clinical decision support. It does not replace independent verification by a licensed provider. It is not FDA-cleared as a clinical-decision-support device — by design.

We rely on the FDA's clinical-decision-support safe-harbor criteria under §520(o)(1)(E) of the Federal Food, Drug, and Cosmetic Act: we cite primary sources, never produce directive output ("give X mg"), and do not analyze medical images or signals. Our output is information that helps a healthcare professional's decision-making — not the decision itself.

The licensed clinician at the bedside is always the decision-maker. NurseMind's role ends at the citation.

If something in the library or in an AI response looks wrong — out of date, misquoted, missing a critical caveat — please tell us:

• From the iOS app: tap the feedback control on any entry or AI response. Reports include the entry ID, the device's anonymous account, and any optional note you add.
• By email: hello@nursemind.app with the entry name and a description of the issue.

Reports route into the same review pipeline used for pre-publication review of high-risk content. Errors involving high-alert drugs, pediatric dosing, or emergency protocols are escalated within 24 hours. Other reports are reviewed within five business days.

We re-pull foundation sources on a quarterly cadence to catch label updates, withdrawn drugs, new boxed warnings, and updated guidelines. Each entry shows a "Last source-fidelity review" date. The library overall was last refreshed on May 9, 2026.

Clinical evidence is a moving target. We are honest about that. If you find an entry that hasn't been refreshed in a long time and the underlying source has changed, tell us — that's exactly what the reporting loop is for.

Questions about how a specific entry was sourced, requests for additional source citations, or concerns about a particular claim? Email hello@nursemind.app. We answer within five business days.

NurseMind, Inc. · Curator model · Last refreshed May 9, 2026.