Surgical AI Training Data  ·  Pre-Seed 2026

The expert signal
surgical robots
cannot generate.

Humynex builds the first cadaver-validated, think-aloud-annotated cognitive-motor training dataset for soft-tissue surgical robotics AI — starting with liposuction, from a surgeon who has performed it ten thousand times.

10k+
Expert procedures
$97B
Surgical AI market by 2032
$500K
Pre-seed raise
0
Competitors with this data

Surgical robotics AI
is starved of expert data.

The surgical robotics market is scaling rapidly. The hardware exists. The AI frameworks exist. The missing layer is training data that captures why an expert surgeon makes each decision — not just what the instrument did.

01

Live OR data is consent-limited and cognitively empty

Hospital recordings are fragmented, de-identification-constrained, and almost never annotated with expert decision logic. What exists is kinematic logs — instrument position and force — without the cognitive layer that explains each micro-decision.

02

Simulation cannot replicate soft tissue

Soft tissue deformation during liposuction — the interaction between cannula pressure, fat planes, fibrous septa, and fluid dynamics — is one of the hardest physical simulation problems in surgical robotics. Domain randomization approaches that work for rigid manipulation break down here.

03

No expert has narrated 10,000 procedures as training data

The operative decision logic of a surgeon with 10,000 repetitions is tacit knowledge. It has never been systematically time-aligned to instrument signals and formatted as supervised learning data. That externalization is what Humynex provides.

"The slow progress toward surgical robot autonomy can be attributed to a few key issues: the scarcity of large, open-source datasets for training, challenges in modeling soft-body deformations encountered during surgeries, and the increased risk of patient injury during clinical trials."

Yip et al. — Science Robotics, 2024

Five synchronized signals.
One expert. Ten thousand cases.

A Humynex cognitive-motor surgical record is a time-indexed bundle of five signal streams captured during expert cadaver demonstration sessions — conducted under willed body program tissue use agreements. No IRB required. No patient consent burden. First session executable within 30 days of facility access.

Video (RGB-D)

Egocentric + field view of the operative site and instrument at 30fps

Force / Pressure

Inline cannula pressure telemetry — the tactile signal driving tissue-plane decisions

Motion (IMU)

3-axis accelerometer & gyroscope on the cannula handle; trajectory and velocity

Intent Narration

Expert surgeon describing decisions in real time — "reducing suction; deep fascia resistance"

Event Tags

Discrete skill-boundary labels: tissue_plane_transition, cannula_swap, abort_pass

// SYNCHRONIZED CAPTURE WINDOW — SINGLE LIPOSUCTION DEMONSTRATION
Video Frames
Motion Telemetry
Force / Pressure
Intent Narration
Event Tags
t₀t₁t₂t₃t₄ →

Unfair advantages
that cannot be replicated.

DM
Photo to be supplied
David L. Matlock, MD, MBA
Co-Founder · Chief Expert Operator

"The operative decision logic of 10,000 procedures — now in machine-readable form."

Dr. Matlock is the primary data asset of Humynex. A board-certified physician and entrepreneur with an MBA from UC Irvine Paul Merage, he has performed over 10,000 liposuction procedures across four decades of practice and trained 435 surgeons in 46 countries in his pioneered techniques.

During Humynex cadaver capture sessions, Dr. Matlock performs expert demonstrations while narrating every operative decision in real time — tissue plane identification, suction pressure adjustment, cannula selection, abort criteria. That narration, synchronized to force and motion telemetry, is the supervisory signal that current surgical robotics AI programs cannot generate from simulation or clinical video.

He is also the President & CEO of Beverly Hills Sunset Surgery Center (1994–Present) and the Laser Vaginal Rejuvenation Institute of America (2003–Present) — two medical businesses he built and continues to operate.

10k+
Expert liposuction procedures Primary data source for cognitive-motor dataset; narrates decisions in real time
435
Surgeons trained globally Laser Vaginal Rejuvenation® · Designer Laser Vaginoplasty® — 46 countries
MD
St. Louis University School of Medicine · ABOG Board Certified Residency: OB/GYN, King Drew Medical Center · California licensed 1979–Present
MBA
UC Irvine Paul Merage School of Business (2000) Physician-founder with formal business training; two operating businesses since 1983
5
Peer-reviewed publications · 2 books · 25+ international presentations Aesthetic Surgery Journal, Journal of Sexual Medicine, Gynecologic Oncology, BJOG

A machine learning scientist serves as technical advisor, providing expertise in multimodal dataset design, synchronization pipelines, and surgical robotics AI model architectures — including imitation learning, diffusion policy, and vision-language-action model fine-tuning. The advisor's near-term role is to define the minimum viable capture schema, write the processing pipeline, and co-author the first technical publication. Advisor identity available under NDA on request.

From expert demonstration
to ML-ready training record.

The Humynex capture platform converts simultaneous sensor streams into structured, machine-learning-ready records through four sequential stages.

Stage 01

Expert Cadaver Demonstration

10,000-case surgeon performs the procedure with real-time think-aloud narration. Willed body program tissue use agreement. No IRB pathway required.

Stage 02

Multimodal Signal Capture

RGB-D camera, IMU on cannula handle, inline pressure transducer, directional microphone. Sub-$3,000 MVP hardware stack. Setup under 30 minutes.

Stage 03

Synchronization Engine

Hardware-level timestamp alignment across all five modalities. Automated quality checks flag drift events before data enters the training pipeline.

Stage 04

Structured Dataset Export

Time-indexed records: video frame + force + motion + narration + event tag, all aligned to the same timestamp. Direct ingestion by standard ML frameworks.

How existing sources compare

Data Source Expert Level Intent / Decision Layer Tissue Validity Humynex Advantage
Public OR video recordings Variable — includes trainees ✕ None ✕ Live patient; consent-limited on all three
Dry-lab phantom studies Mixed ✕ None ✕ Synthetic tissue; poor fidelity on all three
da Vinci kinematic logs ~ Surgeon-level ✕ No narration or decision tags ✕ Clinical; no cadaver control adds cognitive layer
Simulation only N/A ✕ None ✕ Sim-to-real gap unvalidated real tissue + real expert
Humynex Dataset ✓ 10,000-case expert ✓ Full think-aloud + event tags ✓ Willed body cadaver — anatomically valid Unique in market

A $97B market with
no independent data layer.

The AI in robot-assisted surgery segment is growing at 44% CAGR. Intuitive Surgical's proprietary dataset is platform-locked and not commercially available. No independent company has published an expert-annotated, cadaver-validated surgical cognitive-motor dataset.

$97B
AI in surgical robotics by 2032

From $5.5B in 2024, growing at 44.3% CAGR — driven by AI-guided autonomy, intraoperative decision support, and training data platforms. Source: Fortune Business Insights, 2024.

2.68M
da Vinci procedures in 2024 alone

Intuitive Surgical performed 2.68M procedures in 2024 — 17M total. 2,000+ U.S. hospitals operate robotic surgical systems. The hardware infrastructure is already deployed at scale.

$0
Independent expert surgical datasets available

The data infrastructure layer is unoccupied. No independent company has built an open, expert-annotated, cadaver-validated surgical cognitive-motor dataset. That is the gap Humynex fills.

Revenue pathways

Dataset Licensing — OEMs

Intuitive Surgical, Medtronic, Stryker, J&J (Ottava) — all developing next-generation AI-guided platforms that require expert decision-layer training data.

Academic Research Partnerships

Stanford CHARM, JHU LCSR, CMU Biorobotics — established surgical robotics labs with active need for annotated training data and limited access to expert operators.

Government Grants (Non-dilutive)

NIH SBIR Phase I (~$300K), DARPA RSTAS (Robotic Surgical Training and Assessment) — directly aligned program areas with active funding cycles.

Strategic Acquisition

A proprietary expert surgical dataset becomes a strategic acquisition target as OEMs compete on AI differentiation. Intuitive's da Vinci 5 force feedback introduction signals this competitive vector.

Month 1
Capture Protocol & Facility Access
Define ML schema · refine event tag vocabulary · execute willed body program tissue use agreement · procure MVP hardware (<$3K)
Month 2
Cadaver Sessions
5–10 expert liposuction demonstration passes across anatomical variants · real-time think-aloud narration · synchronization validation
Month 3
Dataset & Publication
Process recordings · write 6-page technical note · submit to ISMR 2026 or MICCAI workshop · post to arXiv simultaneously

Let's talk
before the arXiv drops.

Humynex is raising $500K pre-seed. SAFE note, $4M cap, 20% discount. Target close Q2 2026. Seed trigger: published dataset paper + one OEM pilot LOI.

Founder
David L. Matlock, MD, MBA
Company
Humynex Robotics, Inc.
Location
Los Angeles, CA  ·  Beverly Hills
Phone
Stage
Pre-Seed · $500K · SAFE $4M cap · 20% discount
Whitepaper and founder bio available on request. NDA provided for detailed technical disclosure.