Presyndromic
Pulmonary Health
Diagnostics Intelligence
Capturing and Interpreting Sounds and Vibrations Below and Above the Human Threshold of Hearing
WHO WE ARE?
Interdisciplinary team of cutting-edge explainable AI/ML experts; hardware-, software, signal processing, data engineers, ”noise” specialists, & clinicians
WHAT WE DO?
Extract just-in-time actionable health information from data that is currently ignored as “noise”, is unobserved or is uncollectable by current instrumentation for easy, fast, accurate screening and early detection of disease.
WHY?
Presyndromic surveillance for improved detection of emerging public health threats. Predict and detect more threats faster. Stay two steps ahead of threats.
FOR WHO?
Keep the warfighter and America safe!
- A platform of high performance, easy, fast, low-cost point-of-need devices.
- Extract actionable health information from data that is currently ignored as “noise”, is unobserved or is uncollectable by current instrumentation.
- Interoperable solutions that talk to each other, EHR, and other devices.
Game-changing
Presyndromic
Pulmonary Health
Diagnostics Intelligence
imPulseTMUna
Infrasound-to-ultrasound full-bandwidth e-stethoscope passively col- lects bodily audible sounds and inaudible vibroacoustics for interpreta- tion by AI/ML
- Easy, fast, accurate triage to rapidly prioritize casualties for immediate LSIs, such as treatment of significant bleeding and management of airway injuries.
- Gives screening and early detection superpowers to first-responders so the “golden hour” time window for implementing LSIs is not missed or is extended..
imPulseTMTor
Full-bandwidth e-stethoscope plus 4-dry electrode capacitive ECG/EMG/EEG, sPO2, galvanic sensor stack passively collects bodily audible/inaudible vibroacoustics and visible/invisible light biosignal for interpretation by AI/ML
- Non-invasive clinical-grade smart-alarm system that performs a 24/7 deep- and continuous physical.
- Detects biochemical, biomechanical, bioelectric, and/or biophysical function subtle changes impact on biosignal energy, frequency, wave- length fluctuations that are detectable at the skin surface.
- Non-invasive clinical-grade smart-alarm system that performs a 24/7 deep- and continuous physical.
- Detects biochemical, biomechanical, bioelectric, and/or biophysical function subtle changes impact on biosignal energy, frequency, wave- length fluctuations that are detectable at the skin surface.
dCoderTM
AI/ML-driven pre-syndromic pulmonary health screening and early disease detection far-field vibro- acoustic system. Data are collected without contact from 2-to-3 feet. Technology also being tested for “intent” detection to secure US schools.
- Stand-off system to rapidly and autonomously screen and detect hem- orrhage and airway injuries in primary triage
- Context-aware, edge-implemented adaptive –feedback signal process- ing optimization for triage of one or many, e.g., mass casualty incidents (MCIs) in complex settings for resource management.
- Stand-off system to rapidly and autonomously screen and detect hem- orrhage and airway injuries in primary triage
- Context-aware, edge-implemented adaptive –feedback signal process- ing optimization for triage of one or many, e.g., mass casualty incidents (MCIs) in complex settings for resource management.
TrillianTMPatch
Full-bandwidth e-stethoscope plus 4-dry electrode capacitive ECG/EMG/EEG, sPO2, gal- vanic sensor stack passively collects bodily audible/inaudible vibroacoustics and visi- ble/invisible light biosignal for interpretation by AI/ML
- Non-invasive clinical-grade smart-alarm system that performs a 24/7 deep- and continuous physical.
- Detects biochemical, biomechanical, bioelectric, and/or biophysical function subtle changes impact on biosignal energy, frequency, wave- length fluctuations that are detectable at the skin surface.
- Non-invasive clinical-grade smart-alarm system that performs a 24/7 deep- and continuous physical.
- Detects biochemical, biomechanical, bioelectric, and/or biophysical function subtle changes impact on biosignal energy, frequency, wave- length fluctuations that are detectable at the skin surface.
The past: Deep Learning
Billions of nested dot products
Slow to train and query
Opaque
Data-hungry
Prone to unpredictable behaviors
Hard to accommodate domain-specific knowledge
Not compositional
The future: Natural Learning
Concise domain-specific representations
Quick to train and query, adaptable/tunable model
Human-interpretable
Learns from small data
Correct by construction, verifiable and predictable
Ingests examples and domain knowledge
Arbitrarily compositional
Gartner predicts that 70% of organizations will shift their focus from big to small and wide data by 2025. Deep Learning is hitting a wall. Compositionality is the wall. – Gary Marcus, March 2022 [9] I think current deep learning is missing essential pieces. – Yann LeCun, September 2022 [10]
Our algorithms extract actionable health information from novel data that is currently ignored as ‘noise’, is unobserved or is uncollectable by current instrumentation for:
- accurate screening and early detection of disease.
- prediction and detection of health and cyber threats earlier, faster, and more accurately
Smarter & Predictive National & Border Security – Including
- anti-human trafficking, Fraud and Forgery Detection
- Penetrate solid walls to see-through-the-wall and into underground tunnels
- Rapid and precise pandemic and public health response
- Detect and transboundary animal diseases
- Left-of-Boom Directed Energy Weapon Detection and Response
- Proof-of-life immutable digital ID
