Skip to main content

Deterministic 8-primitive biosignal codec with smell/taste multimodal extensions.

Project description

ZPE-Bio


What This Is

Deterministic biosignal compression where reproducibility is mandatory. ECG round-trip fidelity validated against MIT-BIH, PTB-XL, EDB, and NSTDB records — dual Rust and Python codec with embedded reference builds.

For medical-device firmware teams and clinical-data infrastructure engineers: this is the only lane in the family with both a Rust crate and Python package targeting the same signal domain, plus an embedded reference path. The proof lineage is auditable but the release surface is not green. Bio Wearable is NO_GO — its closure bundles are retained for traceability, not treated as release proof.

Readiness: private-stage (2026-03-09). Not a public release packet. Not a clean green-verification snapshot. Historical validation artifacts preserve host-specific paths (lineage, not path authority).

Part of the Zer0pa family. Platform layer: ZPE-IMC.

Field Value
Architecture BIO_WAVEFORM
Encoding ECG_DELTA_V1

Key Metrics

Metric Value Baseline
MIT-BIH 48/48
PTB-XL 1.58× 100/100 entries
EDB_SNR 52.47 dB
NSTDB_SNR 60.49 dB

Source: validation/results/BENCHMARK_SUMMARY.md | validation/results/ptbxl/summary.json | validation/results/edb/summary.json | validation/results/nstdb/summary.json

Competitive Benchmarks

Tool MIT-BIH CR Notes
ZPE-Bio 1.32× Deterministic, bit-identical domain-aware replay
gzip 2.13× Higher CR but no deterministic integrity guarantee

ZPE-Bio targets deterministic integrity, not compression ratio. Gzip achieves higher compression (2.13× vs 1.32×) on MIT-BIH data but does not guarantee bit-identical domain-aware replay. The commercial wedge is determinism + integrity, not raw ratio.

What We Prove

Auditable guarantees backed by committed proof artifacts. Start at AUDITOR_PLAYBOOK.md.

  • Deterministic round-trip fidelity on ECG waveforms
  • Dual implementation: Rust core crate and Python package
  • Wave-1 and Wave-2 readiness artifacts committed
  • IMC contract-consumption alignment verified

What We Don't Claim

  • No claim of fresh green release commit
  • No claim of public release readiness
  • No claim of Bio Wearable validation (NO_GO)
  • No claim of regulatory or FDA compliance
  • No claim of NaN-safe input handling — NaN inputs produce silently incorrect reconstruction (PRD=0.0 reported). This is a known limitation with implications for clinical/regulated deployment. Input validation is the caller's responsibility.

Commercial Readiness

Field Value
Verdict PRIVATE_STAGE
Commit SHA 83dc91685284
Confidence 100% (MIT-BIH integrity passes)
Source validation/results/BENCHMARK_SUMMARY.md

Evaluators: pip install -e . in a clean venv, then run pytest. Contact hello@zer0pa.com for integration guidance.

Tests and Verification

Code Check Verdict
V_01 MIT-BIH integrity passes (48/48) PASS
V_02 PTB-XL benchmark summary committed PASS
V_03 NSTDB benchmark summary committed PASS
V_04 EDB benchmark summary committed PASS
V_05 Sleep-EDF single-file benchmark committed PASS
V_06 Bio Wearable gate FAIL

Proof Anchors

Path State
validation/results/BENCHMARK_SUMMARY.md VERIFIED
validation/results/ptbxl/summary.json VERIFIED
validation/results/nstdb/summary.json VERIFIED
validation/results/edb/summary.json VERIFIED
validation/results/sleep-edfx/summary.json VERIFIED
docs/family/BIO_IMC_ALIGNMENT_REPORT.md VERIFIED

Repo Shape

Field Value
Proof Anchors 6
Modality Lanes 2
Authority Source validation/results/BENCHMARK_SUMMARY.md

ZPE-Bio is the biosignal sector repository for Zero-Point Encoding. It packages a deterministic 8-primitive biosignal codec, a Rust-backed core codec crate, and Bio-specific validation artifacts for Wave-1 and Wave-2 execution.

This repository is a private staging surface as of 2026-03-09. It is not a public release packet and it is not a clean green-verification snapshot.

Quick Start

python -m venv .venv
source .venv/bin/activate
python -m pip install -e ".[dev]"
python -m zpe_bio --help
python -m zpe_bio roundtrip --mode clinical --samples 250

For ECG validation commands:

python -m pip install -e ".[validation]"
python -m zpe_bio encode-ecg --record-id 100 --samples 1000 --json

Optional EEG support requires extra packages and local dataset acquisition:

python -m pip install -e ".[validation,bioeeg]"

Ecosystem

  • Platform-layer contract surface: ZPE-IMC
  • Organization surface: Zer0pa
  • Bio-family alignment artifacts in this repo: docs/family/BIO_IMC_ALIGNMENT_REPORT.md

Who This Is For

Ideal first buyer Medical-device firmware team or clinical-data infrastructure team evaluating deterministic biosignal encoding
Pain statement Biosignal pipelines require reproducibility and auditability — generic compressors are non-deterministic or domain-agnostic
Deployment model Python package + Rust crate, private staged
Family position Staged validation lane — proves the architecture extends to regulated biosignal domains

Current Reality

  • Runtime/package surface exists under python/zpe_bio/.
  • Native codec surface exists under core/rust/.
  • Embedded references exist under embedded/.
  • Proof and readiness artifacts exist under validation/results/ and validation/runbooks/.
  • Family alignment artifacts exist under docs/family/.
  • The source tree contains runnable Python and Rust codec surfaces.
  • The source tree contains committed Wave-1 and Wave-2 readiness artifacts.
  • The source tree contains IMC contract-consumption artifacts for family alignment.

Current Status Snapshot

  • Wave-1 repo substance is real.
  • Wave-2 execution artifacts are present but mixed.
  • Bio Wearable remains NO_GO; its closure bundles are retained for traceability, not treated as release proof.
  • Historical validation artifacts preserve host-specific paths and should be read as lineage, not as current path authority.
  • It does not prove a fresh green release commit.
  • It does not prove public-release readiness.
  • It does not prove Bio Wearable closure.
  • It does not prove that every historical validation artifact path has been normalized.

Repository Map

  • python/zpe_bio/: Python package and CLI
  • core/rust/: Rust codec crate
  • embedded/: embedded reference builds
  • tests/: maintained pytest suite
  • scripts/: operator scripts and generators
  • validation/results/: committed proof and readiness artifacts
  • validation/runbooks/: execution runbooks
  • docs/: repo docs, family alignment, and regulatory material

Audit And Boundaries

  • Proof index: PROOF_INDEX.md
  • Short audit path: AUDITOR_PLAYBOOK.md
  • Public/operator limits: PUBLIC_AUDIT_LIMITS.md
  • Docs landing: docs/README.md
  • Legal boundary summary: docs/LEGAL_BOUNDARIES.md

Open Contradictions

  • Historical GO prose exists, but this staged repo still carries unresolved correctness work.
  • ruff and multimodal manifest verification were not re-cleared in this phase.
  • Regulatory and startup documents contained historical absolute-path references (scrubbed 2026-04-14; now repo-relative).

Read this repo as a private staged baseline for Phase 4.5 and Phase 5, not as a release verdict.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

zpe_bio-0.2.0.tar.gz (82.3 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

zpe_bio-0.2.0-py3-none-any.whl (97.4 kB view details)

Uploaded Python 3

File details

Details for the file zpe_bio-0.2.0.tar.gz.

File metadata

  • Download URL: zpe_bio-0.2.0.tar.gz
  • Upload date:
  • Size: 82.3 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for zpe_bio-0.2.0.tar.gz
Algorithm Hash digest
SHA256 ed97f3b927fb3e77dcf0372ca1ce19a6610488e23ead4bba607b4372888e05be
MD5 ae0e6a5fa4cf546752be90a0611ed698
BLAKE2b-256 9921451b391b1a1ca04a13b8b1070b9189da574688caf77e0453a6b3dcc75161

See more details on using hashes here.

File details

Details for the file zpe_bio-0.2.0-py3-none-any.whl.

File metadata

  • Download URL: zpe_bio-0.2.0-py3-none-any.whl
  • Upload date:
  • Size: 97.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for zpe_bio-0.2.0-py3-none-any.whl
Algorithm Hash digest
SHA256 613c1aa011df88663c8651c40b8a4ce8f19f27ffaf421c63a2b5d7acf92e6b7c
MD5 fc85e8618324a0c45c619a19270b7bf8
BLAKE2b-256 26643f768687c3e0ac810d065cd4bb277d3e940642948ed8bcb8037f9b56fd53

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page