Health Data Vs Ordinary Data

slug: health_data_vs_ordinary_data element_type: TERM mutability: MUTABLE inline: true current_version: 0 status: seed-draft contentURI: null

@health-data is the class of patient information that carries the patient's body, history, vulnerability, and intimate facts. It is not "data" in the generic governance sense; it inherits the more demanding standard of patient-physician confidentiality, applied uniformly across every system that touches it — clinical, AI training, secondary research, hospital analytics, downstream products.

This term distinguishes health data from ordinary data along four operational axes:

1. Sensitivity asymmetry. A leak of ordinary data is reversible (passwords change, cards are reissued). A leak of health data — HIV status, mental-health history, genetic risk, abortion history — is permanent and follows the patient through employment, insurance, family relationships.
2. Re-identification floor. "Anonymized" health data, especially genomic + longitudinal, is increasingly re-identifiable in modern AI contexts. Anonymization alone is not adequate governance.
3. Consent inheritance. Consent for one use does not inherit to another. Consent for clinical care is not consent for training. Consent for one study is not consent for downstream model fine-tuning.
4. Substrate continuity. As the substrate of experience becomes addressable (BCI, neurotech), neural data joins the health-data class. The term must be future-proof against that expansion.

Status

seed-draft · current_version: 0. Placeholder. The legal definitions of "health data" / "PHI" / "sensitive personal data" vary across jurisdictions (HIPAA, GDPR, KVKK, etc.); this term is the federation-internal definition that sits above any single jurisdiction.

Why a federation-level term, not just a legal one

Jurisdictional definitions are the floor, not the ceiling. The federation needs a definition that:

• Holds across jurisdictions (a tool deployed in 50 countries cannot use 50 different floors)
• Anticipates substrate expansion (neural data, behavioral biometrics, ambient health monitoring)
• Is actionable — vendors, hospitals, and regulators can apply it to a specific data flow

Reasoning trail

• Aligns with classical Hippocratic confidentiality, generalized to the AI substrate.
• Aligns with 02-patient-data-sovereignty PRINCIPLE — this term defines what the principle applies to.
• Aligns with 2025–2026 emerging consensus that "data security" in medical AI requires more than HIPAA-style minimums.
• Aligns with UNESCO Recommendation on the Ethics of Neurotechnology (Nov 2025) on neural-data status.

Open questions for the domain author

• Where is the boundary between health data and behavior data (steps, sleep, pulse from a wearable)? Is wearable data health data once it enters a clinical pathway?
• How does the term apply to inferred health data — a model's prediction of disease risk derived from non-clinical inputs?
• What is the consent posture for health data about a person held by another person (genetic relatives, household members in a wearable network)?

Related elements

• 02-patient-data-sovereignty (PRINCIPLE) — the principle this term is the operational substrate of
• re-identification-risk-floor (RULE) — minimum protection against re-identification attacks
• informed-consent-architecture (RULE) — operational consent standard for this data class
• data-flow-audit-mandate (RULE) — what auditing is required

Awaiting domain author

Open the conversation at leviathan.life/forum/medicine.