Skip to content

aaronukgarcia/05-ECG-1000LEAD

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

Status: Pre-Prototype Funding: Seeking £15M Electrodes: 700

⚡ HD-ECGI

High-Density Electrocardiographic Imaging System

Bringing electrophysiology lab precision to the back of an ambulance.

No CT scanner. No hospital. Just a vest and 90 seconds.


🎯 The Problem

Every year, 30,000 people in the UK die from sudden cardiac arrest before reaching hospital. Paramedics have one tool: the 12-lead ECG — technology essentially unchanged since 1942.

Meanwhile, in hospital EP labs, cardiologists use 252-electrode systems with CT-derived heart geometry to map cardiac electrical activity in stunning 3D detail.

The gap is fatal.

💡 The Insight

What if you didn't need CT?

Our hypothesis: 700 electrodes + statistical shape modelling = CT-free geometry estimation accurate enough for pre-hospital use.

If we're right, this unlocks portable EP-grade cardiac mapping for the first time in history.

If we're wrong, we pivot to hospital-based systems and compete on electrode density.

🔬 Technical Specs

Parameter Target Evidence Level
Electrode Count 700 Design spec
Sampling Rate 4000 Hz/channel Validated
CT-free Accuracy ±20mm Hypothesis
Deployment Time <120 seconds Requires validation
Weight <5 kg Mass budget complete
Battery 4 hours continuous 15W power budget

📊 What Success Looks Like

┌─────────────────────────────────────────────────────────────┐
│  SCENARIO              NPV        PROBABILITY    EXPECTED   │
├─────────────────────────────────────────────────────────────┤
│  Full Success         £42M           15%         +£6.3M     │
│  Hospital Only        £18M           25%         +£4.5M     │
│  CT-Based Pivot        £5M           30%         +£1.5M     │
│  Component Sale       -£8M           10%         -£0.8M     │
│  Total Failure       -£15M           20%         -£3.0M     │
├─────────────────────────────────────────────────────────────┤
│  EXPECTED NPV                       100%         +£1.3M     │
└─────────────────────────────────────────────────────────────┘

Break-even: 8% success probability. We estimate 15%.

🚦 Development Gates

Month Gate Criterion If Fail
3 Benchtop PoC 64-ch optical mux <50μs skew TERMINATE
6 Optical Engineer Primary or backup secured PAUSE
9 CT Data Access ≥300 scans for training PAUSE
12 Market Validation N≥50 WTP study positive PIVOT REVIEW
18 CT-Free Algorithm ±30mm accuracy (N≥100) PIVOT

We kill fast. We pivot faster.

🛡️ What We're Not

  • Not a diagnosis machine — Clinical Decision Support only
  • Not autonomous — Human-in-loop required
  • Not validated — All performance figures are engineering targets
  • Not built — This is a pre-prototype investment proposal

📁 Repository Contents

HD-ECGI/
├── docs/
│   └── HD-ECGI-Whitepaper-v6.docx    # Full technical proposal
├── models/
│   └── [placeholder]                  # SSM training (Phase 1)
├── firmware/
│   └── [placeholder]                  # Optical mux control (Phase 1)
├── algorithms/
│   └── [placeholder]                  # CT-free estimation (Phase 1)
└── README.md

🤝 Get Involved

Investors: Read the whitepaper. Grill the assumptions. Fund the hypothesis test.

Optical Engineers: We need you. £90-120K + equity. Contact us.

EP Cardiologists: Clinical advisory board forming. Shape the product.

NHS Trusts: Interested in clinical validation partnership? Let's talk PACS access.

📄 Documentation

Document Description
Technical Whitepaper Full investment memorandum with evidence classification
Liability Matrix Who's responsible when things go wrong
Demographic Validation How we're addressing training data bias

⚠️ Honest Risks

  1. CT-free may not work. Schulze 2019 achieved ±18mm with 12 leads. Extrapolating to 700 is conjecture.

  2. Optical multiplexing at scale is unproven. 66% timing margin, but three components need PoC.

  3. Market validation is weak. N=12 interviews is anecdote, not evidence. Phase 1 fixes this.

  4. Demographic bias could kill. Training data underrepresents South Asian (4×) and obese (4×) populations.

  5. If CT-free fails, we compete against CardioInsight's 10-year head start.

We document these risks because pretending they don't exist is how projects fail.


The best time to map a heart attack was in the EP lab.
The second best time is in the ambulance.


📬 Contact

Aaron Garcia
aaron@garcia.ltd

Version 6.0 | December 2025 | Classification: CONFIDENTIAL

Releases

No releases published

Packages

 
 
 

Contributors