Back to Investors
Intuitive Pathos — Pathos Vigil Seed Deck
Request Conversation
Intuitive Pathos LLC  ·  Central New York
Pathos Vigil
In the ICU, every vital sign is monitored continuously — except the most important one. We built the one that isn't.
Raising
$3M seed  on $15M post
Stage
Working prototype · pre-clearance
Confidential
August 2026
The problem
Level of consciousness is the only primary vital sign
still measured by hand.
HEART RATE CONTINUOUS BLOOD PRESSURE CONTINUOUS SpO₂ CONTINUOUS RESPIRATORY RATE CONTINUOUS TEMPERATURE CONTINUOUS CONSCIOUSNESS BY HAND, q2–4h
Bedside monitors sample at 1 Hz or faster. A RASS or GCS assessment lands roughly every three hours — about 0.0001 Hz — and it is a human judgment call, not a measurement.
Pathos Vigil
02 / 14
What the gap costs
Deterioration doesn't schedule itself
around assessment times.
2–4 h
A repeating blind window
An event that begins ten minutes after a check has hours to progress before anyone looks again — on the most sensitive indicator of neurological deterioration there is.
Subjective
No objective referent
RASS is a human judgment call. The same patient scores differently between two nurses, and between the same nurse at hour 2 and hour 10 of a shift.
Inherited
Early warning has the same gap
NEWS2, MEWS and qSOFA are computed from charted vitals — so the early-warning layer fires on the same episodic cadence, not continuously.
And underneath all three
Population thresholds
miss the individual.
A score that is alarming for one patient is another patient's baseline. Every scoring system in routine use compares a patient to a population. Pathos Vigil compares each patient to themselves — which is simultaneously the clinical argument and the equity argument, in one design decision.
Pathos Vigil
03 / 14
Why it isn't already solved
Continuous monitoring exists. It requires
electrodes on the patient's head.
The incumbent option
EEG and BIS
  • Adhesive electrodes and per-patient consumables
  • Trained placement and ongoing signal maintenance
  • Poorly tolerated by awake or agitated patients — exactly the population that most needs watching
  • Cost and workload that confine it to specific indications
The result: continuous consciousness monitoring is available in principle and almost never used in practice.
What we use instead
The camera already in the room
  • Nothing attached to the patient — no leads, no probes, no consumables
  • No new bedside workflow for nursing to absorb
  • Works on awake, agitated and sedated patients alike
  • Marginal cost per bed-day approaches zero once installed
The hard part was never the optics. It was building a clinical-grade signal out of them — and the regulatory spine to deploy it.
Pathos Vigil
04 / 14
The system
Standard camera in. Clinical metrics out.
All of it on the device.
INPUTS Standard camera RGB, off-the-shelf Microphone Vocalisation, speech response Existing monitor feed HL7 v2.x · HR, BP, SpO₂, RR, temp ON-DEVICE · NVIDIA JETSON Multi-signal neural network Within-subject baseline · each patient referenced to their own rolling normal Demographics deliberately excluded from model input IMAGERY PROCESSED AND DISCARDED BEDSIDE DISPLAY — METRICS ONLY LEVEL OF CONSCIOUSNESS RASS −1 continuous · AVPU · agitation index EARLY WARNING NEWS2 4 continuous · MEWS · qSOFA TREND STATUS Watch — trending NO VIDEO FEED · NO FACE OVERLAY · NO IMAGE STORED OR TRANSMITTED
1 Hz continuous No patient contact On-device inference — no cloud round-trip HL7 v2.x into existing EMR Commodity Jetson hardware
Deployed inference runs as a conventional neural network on Jetson-class GPU hardware. The deployable model is roughly 806K parameters — under 1% of its 1 Hz compute budget — so the bedside unit is commodity silicon, not a research rig.
Pathos Vigil
05 / 14
Architecture as strategy
Two decisions made early that are
very hard to retrofit later.
Decision one
Every patient is their own control
Scores are computed as deviation from that individual's rolling baseline rather than against a population threshold. Demographic variables are kept out of the model input entirely.
PATIENT'S OWN BASELINE DEVIATION SAME ABSOLUTE VALUE, DIFFERENT MEANING PER PATIENT
Why it matters commercially: it is the accuracy argument and the algorithmic-equity argument at once — and FDA increasingly asks for the second.
Decision two
The bedside shows metrics, never video
Imaging is processed and discarded on the device. The operator panel renders scores, trends and alarm banners only. This is an architectural boundary enforced in code — not a policy promise.
CAMERA Frame captured in volatile memory INFERENCE Features only frame destroyed DISPLAY Scores only no image path
Why it matters commercially: it removes the video-retention objection that stalls camera products in hospital privacy and security review — the usual reason these pilots die.
Pathos Vigil
06 / 14
Honest status
What is true today — and what isn't.
Every number in this deck is one we can source. Here is the part most decks leave out.
Built and working
  • Working end-to-end prototype running on Jetson bedside hardware
  • FDA Pre-Submission package drafted
  • IEC 62304 Class C development under an emerging ISO 13485-aligned quality system
  • ISO 14971 risk file with FMEA, INCOSE/ISO 15288 systems-engineering package, SOUP bill of materials, PCCP
  • Requirements → architecture → verification traceability, audit-gated on every commit
Not yet true
  • Not FDA cleared. The Pre-Submission has been drafted, never submitted
  • Not clinically validated. No prospective clinical evidence exists yet
  • The deep model is supervised on synthetic development labels
  • Our one retrain on real data (CinC-2019 vitals) scored below NEWS2 on discrimination
  • No superiority to any existing score or device is claimed — by us, anywhere
Why we lead with this
Class C medical software succeeds or fails on the habit of never letting an unmeasured value render as a real one. We enforce that in the codebase with an automated gate that blocks the commit. That discipline is very hard to install after the fact — and it is what makes the evidence we generate next worth believing.
Pathos Vigil
07 / 14
The evidence plan
A prospective study designed to be able
to prove us wrong.
NIH SBIR Phase I · R43 NINR September 5, 2026 receipt Prospective · observational · single site
Aim 1
Consciousness
Continuous RASS estimation validated against paired expert bedside assessment as the reference standard.
weighted Cohen's κ
mean absolute error
Aim 2
Deterioration
AUROC at 30 / 60 / 120 / 240 minutes of lead time against chart-confirmed events — rapid response, unplanned intubation, arrest, 24 h ICU readmission.
vs continuously computed
NEWS2 · MEWS · qSOFA
Aim 3
Subgroups
Accuracy across patient subgroups — the direct test of whether referencing each patient to their own baseline delivers the equity property it is designed for.
demographics excluded
from model input
The asset this creates
A dataset that
does not exist
Paired continuous video and audio with expert bedside RASS assessment is not available for purchase from anyone, at any price. Whoever collects it first holds the training and validation substrate for this entire category. The study is simultaneously our evidence and our moat.
Pathos Vigil
08 / 14
Customer and market
Sold per bed, to the Chief Nursing Officer.
Who signs
The CNO is the right door
This reduces a surveillance and documentation burden that falls squarely on nursing. The Chief Nursing Officer is the economic buyer, the Critical Care Medical Director is the clinical champion, and purchase routes through the hospital's value-analysis committee.
A nurse-founder walks into that room as a peer describing a problem they personally worked around. That is a durable access advantage, and it is not one a competitor can hire for quickly.
Beachhead
Regional systems in Central & Western New York
Close enough to support directly and under exactly the staffing pressure this addresses. Large academic centres are engaged first as validation and publication partners — which is how the CNO audience is reached at scale.
Where the beds are
SettingUS bedsList / bed / mo
Adult ICU & critical care~100,000$500
Step-down / high dependency~200,000$250–400
General med-surg~600,000$100–150
Skilled nursing / assisted living~1,200,000$75–125
ICU alone, at list
~$600M
annual, US adult ICU beds only
Entry wedge
1 unit
land one ICU, expand by floor
Bed counts are approximate US figures used for sizing. The $600M line is list-price arithmetic against ICU beds only — it is a market-size illustration, not a revenue forecast.
Pathos Vigil
09 / 14
Business model
Recurring per-bed licence. The installed
fleet appreciates.
How the unit economics work
~$500
hardware BOM per bed
at volume, amortised 5 yrs
~74%
mature blended
gross margin
The same installed device earns the next tier through a software update — a step-down unit can gain ICU-tier capability without new hardware. Revenue per bed rises without a new sale, which is what makes the every-bedside thesis pay for itself.
Revenue lines
LineModelFDA dependent
Hospital per-bed licenceMonthly, tiered by acuityYes
Veterinary monitoringPer-cage / per-clinic subscriptionNo
Non-clinical wellnessConsumer / employer subscriptionNo
Federal, disaster & pharmaUnit procurement / per-trial contractLater
The two non-FDA-dependent lines run on the identical core at near-zero incremental engineering — the near-term revenue floor that de-risks the runway while the clinical path clears regulatory.
Plan of record — clinical build
~$157M Y5  ·  ~$1.1B Y10
Floor — vet & wellness only
~$40M Y10, no FDA dependency
A structural commitment
A Mercy tier priced at zero — refugee, disaster and LMIC settings — funded by a 5% foundation revenue covenant. A deliberate constraint, not marketing.
Revenue figures are forward-looking management projections dependent on regulatory clearance not yet obtained. They are not guarantees.
Pathos Vigil
10 / 14
Defensibility
Four moats, only one of which
is a patent.
01 — Intellectual property
Inventions under patent counsel
Carson Patents as paid primary counsel plus pro bono counsel through Volunteer Lawyers for the Arts. A Round 1 prior-art search completed May 2026; counsel's professional read is that the foundational claims and the eight-axis architectural framework are novel under 35 U.S.C. §102 against the eighteen cited references.
Counsel's assessment from a prior-art search — not a USPTO determination. Zero applications are currently pending; the §102 priority clock is clean. A first provisional is in preparation.
02 — Data
A corpus nobody can buy
Paired continuous video and audio with expert bedside RASS labels does not exist as a purchasable dataset. It has to be collected prospectively, under IRB, with clinician time attached to every label. The first mover compounds; the second pays the same cost years later.
03 — Architecture
Within-subject baseline
A competitor training on population thresholds cannot retrofit per-patient referencing without rebuilding the model, the validation, and the regulatory submission that rests on both.
04 — Regulatory posture
The boring, decisive one
Class C design controls, a risk file, traceability, configuration management and a metrics-only privacy architecture — built in from the start and kept aligned to both FDA and EU MDR/UKCA, so the same evidence opens two markets. This is 18 months a fast follower cannot compress.
Pathos Vigil
11 / 14
Where we are
Conversations, artifacts and applications —
stated as exactly that.
Clinical path
  • Co-investigator conversation underway with Pulmonary & Critical Care at URMC, routed via UR CTSI
  • Second co-investigator channel open at SUNY Upstate
  • Shadow-mode pilot design and security / integration FAQ already drafted for a regional system
Funding path
  • NIH / NINR SBIR R43 — September 5 receipt date; programme fit call held with NINR
  • NYSTAR innovation matching grant sequenced behind it
  • Active client of TDO, NYSTAR's CNY Technology Development Center, for IEC 62304 V&V
Technology & industry
  • Working bedside build on NVIDIA Jetson; inbound interest from NVIDIA Healthcare
  • CTO-level conversation with IBM on reference architecture
  • Full regulatory artifact set drafted — Pre-Sub, risk file, SE package, traceability
Said plainly
What we do
not have
No executed pilot agreement. No revenue. No clinical validation. No signed co-investigator. These are live conversations and applications in flight, and we describe them that way in every room — including this one. The seed exists to convert precisely this list.
Pathos Vigil
12 / 14
Team
The person who lived the problem
built the solution.
Founder
Matthew Lashomb, RN, BSN
Founder & Principal Investigator
11 yrs
inpatient bedside nursing
7 / 4
years psychiatry / neuro & stroke
Designed and built the entire system — the clinical model, the edge inference path, the HL7 integration and the full regulatory artifact set. The rare part is not the nursing or the engineering separately. It is that the person who lived the workflow gap is the one specifying the software that fills it. Every design decision traces to a shift he worked.
The bench — contracted, not hypothetical
  • IEC 62304 V&V lead engaged through TDO, NYSTAR's CNY Technology Development Center
  • Carson Patents — paid primary patent counsel; Volunteer Lawyers for the Arts — pro bono counsel
  • Clinical co-investigator in active negotiation at URMC and SUNY Upstate
The honest gap — and the first hire
One full-time person is the binding constraint on outreach velocity, and we say so. The first operational hire post-seed is a chief of staff to own scheduling, grant project management and partner follow-up — the actual bottleneck. Embedded-ML and clinical-affairs hires follow the validation study.
Pathos Vigil
13 / 14
The ask
$3M seed on $15M post.
Cleanly capitalised: no prior equity, no notes, no SAFEs. Self-funded to date. We are assembling the syndicate and want a state and strategic co-investor alongside the lead.
Use of funds — three milestones, not three categories
MILESTONE 01
Close the FDA path
Pre-Submission meeting with CDRH, then De Novo preparation with a Predetermined Change Control Plan. Kept aligned to EU MDR/UKCA so one evidence package opens two markets.
MILESTONE 02
Design-freeze the bedside unit
Harden the Jetson build to a sealed, reproducible, verifiable configuration — the baseline the risk file, V&V records and SOUP list all pin to.
MILESTONE 03
First clinical validation
Stand up the prospective study and the first pilot site — converting the honest-status slide, line by line, into evidence.
Stacks with
non-dilutive
NIH / NINR SBIR R43 (Sept 5 receipt) plus a NYSTAR innovation matching grant. Federal and state non-dilutive capital leveraging the equity round rather than competing with it.
Contact
Matthew Lashomb, RN, BSN
Founder & Principal Investigator · Intuitive Pathos LLC
matthew.lashomb@gmail.com
Confidential — Intuitive Pathos LLC
14 / 14