Free Wearable Corpus — Schema v0.1

Purpose

A structured catalog of every wearable and body-measurement device — wristworn, finger, eyewear, ear, head, torso, limb, foot, garment, patch, implantable, ingestible — disclosed in private patents, open source, science fiction (film/TV/comics/games/literature), academic literature, regulatory filings, and standards bodies.

The corpus IS the prior art commons. Every entry, at the moment of timestamped commit and quarterly release, is a defensive publication artifact. prior_art_notes is not editorial commentary — it is the element-by-element analysis a competent examiner or invalidity-contention attorney can cite.

The corpus is a lexicon, not a catalog. A lexicon entry isn’t annotation: it is the demonstration that a specific disclosure anticipates or motivates obviousness toward a specific claim phrase. Retrieval equals citation.

Two structural commitments that differ from the Free Humanoid Corpus

1. Multi-axis tagging — form factor × sensor × algorithm

Wearable claims are component-anchored, not platform-anchored. A claim of “a wristworn device comprising a photoplethysmography sensor and a processor configured to compute heart rate variability” reads on a combination of three independent elements:

• form factor (watch, ring, glasses, patch, …)
• sensor (ppg, ecg, eeg, accelerometer, …)
• algorithm / output (hrv, afib-detection, sleep-staging, …)

The corpus indexes by component, not by combination. Each entry is multi-tagged across all three axes. Cross-cuts are generated per tag and also per tag-intersection where useful. This is what lets the corpus scale to wearable density (1000+ entries) without bottlenecking on editorial.

2. Combination obviousness is template, not per-combination work

Because adding a known sensor to a known wearable form factor is per se obvious under KSR / MPEP § 2143 Rationale A (combining prior art elements according to known methods to yield predictable results), the §103 proof for any (form factor × sensor) tuple is boilerplate that plugs into the template, not bespoke per-combination analysis. See OBVIOUSNESS_TEMPLATE.md.

Entries supply the component disclosures. The template supplies the combination argument. The corpus is component cross-cuts plus one template.

Quality bar for entries

An entry is commons-grade (not draft: true) when:

• disclosure_citation resolves to a primary source verifiable by a third party (paper DOI, patent number, regulatory filing ID, episode ID with first-air-date, standards-document number, archive URL).
• first_disclosure_date is the earliest verifiable public disclosure, defensible against challenge.
• prior_art_notes reads as a 102/103 anticipation analysis someone unrelated to us could cite without rewriting. It identifies specific components disclosed (form factor, sensor modalities, algorithms, output channels) and what claims those disclosures could anticipate.
• sources cite primary references.
• For patented entries, ip_citations lists actual patent numbers.

Entries below the bar may be merged with draft: true to make incremental progress visible.

Entry tiers

Density without editorial bottleneck requires tiered entries.

• Tier 1 (commons-grade, full): complete prior_art_notes, full multi-axis tagging, primary citations. Used as anchor entries in cross-cuts.
• Tier 2 (reference-only): title, date, citation, bucket, minimal tags. Populates the long tail so lookups don’t return empty. Marked tier: 2 and not subject to the full quality bar (still requires a resolvable primary citation).

Both tiers participate in cross-cut generation. Tier 1 entries get full rendering; Tier 2 entries get one-line rendering with date + citation.

License posture

CC0-1.0. The corpus is public domain dedication. Individual entries cite their sources but the structured catalog is unencumbered.

Timestamping

Same ceremony as the Free Humanoid Corpus. Three independent layers per release: FreeTSA (RFC 3161), DigiCert (RFC 3161), OpenTimestamps (Bitcoin-anchored). See TIMESTAMPING.md.

Entry schema

Each entry is one record. Required [R], optional [O].

id                       [R]  slug, kebab-case, globally unique within corpus
canonical_name           [R]  the name most commonly used in literature
aliases                  [O]  list of alternate names, model numbers, codenames
corpus                   [R]  one of: private | open | fictional | academic | regulatory | standards
first_disclosure_date    [R]  ISO 8601 date or year; earliest verifiable public reference
disclosure_citation      [R]  source for the first_disclosure_date (DOI, ISBN, URL, patent #, 510(k) #, episode ID, standards doc #)
creator                  [R]  company / lab / studio / author / collective / agency / SDO
creator_country          [O]  ISO country code of creator

form_factor              [R]  primary form factor; see form-factor taxonomy
form_factor_tags         [O]  additional form-factor tags (multi-form devices)
contact_surface          [O]  skin | scalp | ear | nasal | dental | ocular | sublingual | intra-abdominal | vascular | sub-dermal | textile-mediated | non-contact
anatomical_target        [O]  list of body regions targeted (wrist, temporal-lobe, finger, chest, scalp-Fpz, etc.)

sensors                  [O]  list of sensor tags from sensor taxonomy
sensor_details           [O]  free text — wavelengths, electrode geometry, sampling rate, DR, channel count
actuators                [O]  list of actuator/output tags (haptic-eccentric, haptic-piezo, electrical-stim-tens, optical-stim, audio, display, ...)
compute                  [O]  on-device chip / MCU / accelerator, edge-vs-cloud split
connectivity             [O]  ble | wifi | nfc | lte-m | thread | wired | sneakernet | none
power_source             [O]  cr2032 | li-ion-rechargeable | nfc-harvested | piezoharvest | tethered | fictional | unknown
runtime                  [O]  hours or days

algorithms               [O]  list of algorithm tags (hrv, afib-detection, sleep-staging, fall-detection, seizure-detection, ...)
output_modalities        [O]  haptic | visual-display | audio | electrical-stim | thermal | drug-delivery | data-only
clinical_endpoints       [O]  list of claimed/disclosed clinical outputs (spo2, glucose, blood-pressure, eeg-bandpower, ...)

ip_status                [R]  patented | open-permissive | open-copyleft | public-domain | fictional | regulatory-filing | standards | trade-secret | unknown
ip_citations             [O]  patent numbers, license identifiers, 510(k) numbers, NCT trial IDs, repo URLs, standards doc IDs

regulatory_pathway       [O]  fda-510k | fda-de-novo | fda-pma | ce-mdr | ce-ivdr | exempt | none | fictional
regulatory_predicates    [O]  list of predicate device IDs (for 510(k) entries)
clinical_trials          [O]  list of NCT/ISRCTN identifiers

prior_art_notes          [O]  element-by-element 102/103 analysis; what claims this entry could anticipate
lineage_ancestors        [O]  ids of entries this design descends from
lineage_descendants      [O]  ids of entries that descend from this (filled in later passes)

sources                  [R]  list of citations (papers, books, articles, repos, episodes, filings, standards)
cpc_classifications      [O]  CPC codes for examiner discoverability (e.g., A61B 5/02416 for PPG)
notes                    [O]  free-text observations
draft                    [O]  boolean; true if the entry has not yet cleared the quality bar
tier                     [O]  integer 1 (full, default) or 2 (reference-only)
schema_version           [R]  integer matching schema spec version
last_updated             [R]  ISO 8601 date

Form-factor taxonomy

The primary axis. Patent claims tend to read on a specific form factor. Fiction’s wearable coverage is densest in watch, bracelet, and glasses.

• watch — wrist-mounted device with display
• bracelet — wrist-mounted device without display (or jewelry-style)
• ring — finger-worn
• glasses — eyewear, with or without display / HUD
• goggles — fully enclosed eyewear, includes VR/AR HMDs
• contact-lens — corneal contact
• earbud — in-ear
• over-ear-headphone — over-ear / on-ear with sensors
• hearing-aid — behind-the-ear / in-canal hearing aid form factor
• headband — forehead / scalp band
• cap — full-head cap (EEG cap, helmet)
• helmet — protective helmet with embedded electronics
• patch — adhesive skin patch
• garment — clothing-integrated (shirt, sock, glove, sports bra, smart-fabric)
• belt — waist-mounted
• pendant — neck-worn
• armband — upper-arm band
• legband — leg or thigh band
• sock — foot-worn textile
• insole — shoe insert
• shoe — full footwear
• body-camera — chest-mounted camera
• exoskeleton — wearable exoskeleton
• implantable — surgically implanted
• ingestible — swallowed sensor
• dental — tooth-mounted / mouthguard
• tattoo-electronic — epidermal electronics, sub-dermal printed
• fictional-other — fiction with no real-world form-factor match
• other — escape hatch with explanation in notes

Sensor taxonomy

Tags applied to sensors. An entry tags every sensor modality it discloses.

Optical / photonic

• sensor-ppg — photoplethysmography (HR, HRV, SpO2-base)
• sensor-spo2 — pulse oximetry (red + IR)
• sensor-multi-wavelength-ppg — beyond red/IR, e.g. green/blue/NIR
• sensor-fnirs — functional near-infrared spectroscopy
• sensor-raman — Raman spectroscopy
• sensor-optical-glucose — optical glucose sensing (transcutaneous)
• sensor-camera-rgb — visible camera
• sensor-camera-ir — infrared camera
• sensor-camera-thermal — thermal imager
• sensor-camera-eye — eye-tracking camera
• sensor-photodiode-ambient — ambient light

Electrophysiological

• sensor-ecg — electrocardiography
• sensor-eeg — electroencephalography
• sensor-emg — electromyography
• sensor-eog — electrooculography
• sensor-gsr — galvanic skin response / EDA
• sensor-bioimpedance — body composition / hydration
• sensor-respiration-impedance — impedance pneumography

Mechanical / motion

• sensor-accelerometer — 3-axis accel
• sensor-gyroscope — 3-axis gyro
• sensor-magnetometer — 3-axis mag
• sensor-barometer — pressure altimeter
• sensor-piezoelectric — piezo strain / vibration
• sensor-strain-gauge — strain
• sensor-pressure-skin — skin-pressure / cuff pressure

Acoustic

• sensor-microphone-air — ambient audio
• sensor-microphone-bone — bone-conduction mic
• sensor-stethoscope-digital — digital auscultation
• sensor-ppg-acoustic — acoustic-PPG variants

Biochemical

• sensor-glucose-cgm — continuous glucose monitor (electrochemical)
• sensor-lactate — sweat / interstitial lactate
• sensor-cortisol — sweat / saliva cortisol
• sensor-electrolyte — Na/K/Cl in sweat
• sensor-ph — pH in sweat / interstitial
• sensor-ammonia — ammonia sensing
• sensor-alcohol-transdermal — transdermal alcohol
• sensor-ketone — ketone sensing
• sensor-uric-acid — uric acid
• sensor-creatinine — creatinine
• sensor-cytokine — cytokine (IL-6 etc.) sensing
• sensor-dna-onchip — on-device DNA / nucleic-acid sensing

Thermal

• sensor-skin-temperature — skin temp
• sensor-core-temperature — core temp (ingestible / tympanic)
• sensor-heat-flux — heat flux

Hydration / sweat

• sensor-sweat-rate — sweat rate
• sensor-microfluidic-sweat-collection — sweat collection channel

Hemodynamic / vascular

• sensor-cuffless-bp-ptt — pulse-transit-time cuffless BP
• sensor-cuffless-bp-tonometry — applanation tonometry
• sensor-cuffless-bp-volume-clamp — finger volume clamp
• sensor-arterial-stiffness — arterial stiffness inference

Neural / brain

• sensor-dry-eeg-electrode — dry contact EEG electrode
• sensor-saline-eeg-electrode — wet/saline EEG electrode
• sensor-microneedle-eeg — microneedle EEG electrode
• sensor-meg-mini — miniature MEG (OPM-based)

Other

• sensor-uv — UV exposure
• sensor-tof-distance — time-of-flight distance
• sensor-radar-mmwave — mmWave radar for HR / respiration
• sensor-ultrasound-wearable — wearable ultrasound
• sensor-air-quality — VOC / PM2.5 wearable

Algorithm / output taxonomy

Tags applied to algorithms. An entry tags every algorithm it discloses, whether on-device or cloud.

Cardiovascular

• algo-hr — heart rate
• algo-hrv — heart rate variability
• algo-afib-detection — atrial fibrillation detection
• algo-arrhythmia-classification — multi-class arrhythmia
• algo-pwv-bp-estimation — pulse-wave-velocity blood pressure
• algo-stress-index — autonomic stress index

Sleep

• algo-sleep-staging — REM/NREM/wake staging
• algo-sleep-apnea-detection — apnea-hypopnea index from wearable
• algo-snore-detection

Activity / motion

• algo-step-count
• algo-fall-detection
• algo-gait-analysis
• algo-activity-classification — walking/running/cycling/etc.
• algo-calorie-estimation
• algo-posture-detection
• algo-tremor-detection — Parkinsonian / essential tremor
• algo-bradykinesia-detection

Neural / cognitive

• algo-seizure-detection
• algo-cognitive-workload
• algo-drowsiness-detection
• algo-attention-state
• algo-bci-ssvep
• algo-bci-motor-imagery
• algo-bci-p300
• algo-erp-classification

Metabolic / chemical

• algo-glucose-noninvasive
• algo-glucose-cgm-readout
• algo-hydration-status
• algo-electrolyte-trend

Other

• algo-hand-gesture-emg
• algo-keystroke-emg
• algo-speech-bone-conduction
• algo-eye-gaze-tracking
• algo-pupillometry
• algo-emotion-recognition
• algo-spo2-estimation
• algo-respiratory-rate
• algo-cough-detection
• algo-uv-dose-tracking
• algo-thermoregulation-modeling

IP status definitions

• patented — known patents asserted, with patent numbers
• open-permissive — MIT, BSD, Apache, CERN-OHL-P, etc.
• open-copyleft — GPL, CERN-OHL-S, etc.
• public-domain — CC0 or equivalent dedication, or expired protection
• fictional — exists only in narrative; the engineering description is unencumbered as prior art
• regulatory-filing — disclosed via FDA / CE filing; the filed disclosure is public record
• standards — disclosed via standards body document; document is publicly citable
• trade-secret — claimed but not disclosed
• unknown — needs investigation

Corpus buckets

• private — commercial products with disclosed engineering, including patent-thicket holders. Reference points for what’s been claimed.
• open — open-hardware / open-source wearables. Already public-domain or permissive.
• fictional — film, TV, comics, games, literature. Cited successfully as 102 prior art when specific enough; cleanly serves 103 motivation-to-combine for non-enabling disclosures.
• academic — peer-reviewed papers, theses, conference proceedings. Already prior art automatically by publication.
• regulatory — FDA 510(k) submissions, FDA De Novo / PMA decision summaries, CE MDR/IVDR technical documentation, predicate device chains. Public-record disclosures that examiners often miss.
• standards — IEEE 11073, Bluetooth SIG profiles (Heart Rate, CGMS, Pulse Oximeter, Continuous Glucose), Continua Alliance, ISO/IEC 80601 series, AAMI/ANSI standards drafts. Disclose enabling implementations in normative annexes.

Cross-cut generation

Each entry’s tags drive cross-cut file generation. Three cross-cut families:

1. Single-axis cross-cuts — one file per tag in form_factor/form_factor_tags, sensors, algorithms. The foundational view: every disclosure of sensor-ppg in chronological order, every disclosure of watch form factor in chronological order, etc.
2. Form-factor × sensor intersections — one file per (form-factor, sensor) pair that has ≥3 entries. E.g., watch__sensor-ppg.md, glasses__sensor-eeg.md. These are the direct inputs to the obviousness template.
3. Sensor × algorithm intersections — one file per (sensor, algorithm) pair with ≥3 entries. E.g., sensor-ppg__algo-hrv.md, sensor-eeg__algo-seizure-detection.md.

Intersection cross-cuts are generated automatically; they are not hand-curated. Single-axis cross-cuts may have hand-written narrative preambles on the canonical ones.

Storage format

• Master corpus: corpus.jsonl — one JSON object per line, append-only, git-tracked.
• Per-corpus mirrors: private.jsonl, open.jsonl, fictional.jsonl, academic.jsonl, regulatory.jsonl, standards.jsonl.
• Index: CORPUS_INDEX.md — generated, alphabetical by canonical_name.
• Lineage graph: lineage.json — derived, ancestor/descendant DAG.

Versioning

Schema version increments on breaking changes. Entries carry schema_version so older entries can be migrated. v0.1 is the starting point and should be expected to evolve once seed slices expose inadequacies.