Defensive Publication (Public Domain): Open-Source Diagnostic Braid (OSDB)

Defensive Disclosure • January 2026 • Commons Health Innovation

A modular, composable diagnostic framework that treats health as a braided network of signals — biological, behavioral, and environmental — enabling anyone to build low-cost, privacy-preserving diagnostics without vendor lock-in.

Status: Speculative but fully implementable with today’s off-the-shelf sensors, assays, and open-source tools.

License Intent: Public Domain (CC0 1.0 Universal). No rights reserved. This disclosure is designed to establish prior art and prevent proprietary enclosure of braided diagnostic architectures.

This repository publishes the complete conceptual blueprint, architecture, data model, and reference implementation guidance for the Open-Source Diagnostic Braid (OSDB).

Core Idea: Build a Braid, Not a Single Test

Health states are latent — you cannot measure them directly.

You can only measure many noisy, partial proxies.

Truth emerges from braid tension: patterns of agreement and disagreement across multiple weak signals.

1. The Braid Principle

Each “strand” is an independent measurement channel:

1. Molecular strand – e.g., CRP, ferritin, glucose, lactate, pathogen markers
2. Physiology strand – HR, HRV, SpO₂, temperature, BP, sleep stages
3. Symptom/phenotype strand – structured diary, cough audio, rash photos, pain scales
4. Exposure strand – air quality, allergens, diet log, medications, travel history
5. Microbiome/immune strand – optional slower-changing context
6. Baseline strand – your personal normal (the most powerful and underrated biomarker)

The braid fuses these into:

• A probabilistic risk score
• An uncertainty estimate
• A next-best-measurement suggestion

2. OSDB Architecture (Modular & Buildable)

A. Strand Adapters (Input Standardization)

Convert raw signals into a unified event format:

{
  "timestamp": "2026-01-07T10:32:00Z",
  "strand_type": "physiology",
  "feature_vector": {"hr": 92, "hrv": 28, "spo2": 96},
  "quality_flags": {"motion_artifact": false, "good_contact": true},
  "provenance": "fitbit_ble_export"
}

Supported adapters (example list):

• Phone-camera lateral flow assay readers
• Manual lab panel entry / CSV import
• Wearables (Apple Health, Google Fit, BLE)
• Structured questionnaires (FHIR-compatible)
• Environmental APIs / local sensors

B. Feature Weaving Layer

Primary features are deltas from personal baseline, not population norms.

Examples:

• ΔCRP from user baseline
• 3-day slope of resting HR
• Discordance score (e.g., high temp but normal CRP)

C. Braid Inference Engine (Probabilistic Fusion)

Recommended approach:

• Bayesian graphical model / factor graph over latent health states
• Each strand contributes likelihood factors
• Output: full posterior distribution + explainable evidence chains

Example output:

“Moderate probability of viral syndrome (62%). Evidence (coherence): rising fever trend + mild tachycardia + cough acoustics. Evidence (contradiction): normal SpO₂, no chest pain. Uncertainty drivers: missing CRP, low-quality sleep HRV.”

D. Active Measurement Planner

Information-theoretic selection of next measurement that maximally reduces uncertainty per cost/effort.

Examples:

• “Repeat temperature in 4 hours”
• “Perform rapid strep test”
• “Spot-check SpO₂ after light activity”

E. Safety & Governance Layer (Hard Guardrails)

Non-negotiable rules:

• Immediate escalation for red flags (SpO₂ < 92%, severe chest pain, stroke signs)
• Clear disclaimers: decision support only, never autonomous diagnosis
• “Do not delay professional care” triggers

3. Braid Tension Metrics (The Secret Sauce)

• Coherence: Signals align with a hypothesized state
• Contradiction: Signals conflict → points to artifact or alternative etiology
• Dynamic Strand Weighting: Reliability adjusted per session (lighting, motion, medication effects)

Cheap sensors become useful not by being perfect, but by being honest about their limitations.

4. Minimal Viable Reference Implementation (OSDB-MVP)

Focus: Triage & monitoring of common inflammation/infection/recovery patterns

Low-cost inputs:

• Temperature (oral thermometer)
• Resting HR & HRV (wearable or manual pulse)
• SpO₂ (pulse oximeter)
• Structured symptom checklist
• Optional CRP (home or low-cost lab)

Outputs:

• Risk bands: Low / Watch / Urgent
• Uncertainty estimate + missing data suggestions
• Recovery trajectory tracking

5. Data Model: Personal Baseline First

Population ranges are blunt. OSDB treats each individual as their own control:

• Typical resting HR, HRV, temperature
• Personal CRP/ferritin history (if measured)
• Seasonal patterns, medication signatures

Question shifts from “Are you in range?” → “What changed for you?”

6. Validation Plan (Open & Reproducible)

1. Synthetic validation – simulated trajectories + sensor noise
2. Retrospective – de-identified datasets
3. Prospective – community/clinic studies with clinician oversight

Key metrics:

• Calibration curves
• Decision-curve analysis
• Abstention performance
• Fairness across demographics

7. Anti-Enclosure Design Features

• Open interface spec for all strand adapters
• Explainability by construction (visible graph + weights)
• Offline-first / on-device capable (privacy)
• Federated learning optional
• Full provenance logging

Example Narrative Output

Posterior: Moderate probability viral syndrome (62%); low probability bacterial pneumonia (18%)
Coherence evidence: Fever trend upward + mild tachycardia + cough onset timing
Contradiction evidence: Normal SpO₂, no reported chest pain
Uncertainty drivers: Missing inflammation marker (CRP), low-quality overnight HRV
Next best measurement: Spot-check SpO₂ after 2-minute walk; consider CRP if symptoms worsen
Safety note: If SpO₂ drops below 92% or breathlessness increases → seek urgent care immediately

This is braided diagnostics: not false certainty, but guided, inspectable clarity.