EDGE-AI IP
Fully Integrated and Optimized Edge AI
Mastering Edge-AI demands cutting-edge technology and expertise to overcome lingering energy consumption and computational hurdles.
ALGORITHMS
Grow-and-Prune
CTRL
TUTOR
A standout feature of NeuTigers' technology is its incremental learning framework, which efficiently updates neural networks with new data. Inspired by the human brain's remarkable ability to learn and adapt, Grow-and-Prune mimics the natural process of neural network development.
CTRL spots labeling errors, making training data more accurate. Our TUTOR framework iteratively combines real and synthetic data to produce the most accurate DNNs.
NeuTigers’ Grow-and-Prune technology, combined with CTRL and TUTOR is a paradigm shift in AI, offering more efficient, compute- and size-optimized edge-DNNs, which are vital for applications requiring quick inference, like autonomous disease detection, driving and IoT devices.
SCouT
SCouT, a proprietary analytics framework, uses advanced AI to analyze complex (spatiotemporal) health data over time to recreate time- and subject-efficient clinical trials.
Our approach means it takes fewer patients and less time to arrive to the conclusion of a trial with the same statistical power as current trial methodology.
SCouT can also enable precise, data-driven predictions and decisions in public health policy, and individual patient treatment, ultimately leading to better health outcomes and more informed healthcare strategies.
APPLICATIONS
DiabDeep
Pervasive Diabetes Diagnosis based on Wearable Medical Sensors and Efficient Neural Networks
Diabetes, one of the top 10 causes of death globally, is a major cause of blindness, kidney failure, heart attacks, stroke and other significant medical complications.
The NeuTigers teams is conducting a 2-year study to develop DiabDeep, a wearables-based diabetes detection application to make it easy for patients anywhere in the world to find out if they are diabetic and to accelerate the process of managing the disease.
We believe DiabDeep has the potential to save countless lives and to relieve the economic and resource burden on healthcare systems across the world. These are the outcomes that inspire us and drive us every day
CovidDeep
COVID-19 screening using a smartwatch in minutes with
90%+ accuracy.
The COVID-19 pandemic posed a major threat that shut down the global economy for over a year. One of the greatest challenges in containing the spread of this highly infectious disease, was to quickly identify and rapidly isolate confirmed cases and their contacts.
In response to the global COVID-19 pandemic, NeuTigers developed a non-invasive and rapidly scalable solution leveraging commercially available wearable medical sensors (WMSs) for the pervasive diagnosis of infection.
The final deliverable, CovidDeep, is a prediction smartphone app that embeds the neural network developed using data captured during a 2,000-patient controlled prospective study in Europe. CovidDeep provides a 97% detection accuracy (twice as effective as current triage tools such as temperature checks and questionnaires), can detect asymptomatic infections, and is currently under regulatory review for approval in two different countries.
Generative AI
We are continuously seeking out new technologies to augment the impact of the solutions we deliver. With this in mind, we are currently working on the integration of a generative AI engine into our offerings.
The goal is to leverage advanced language processing abilities to deliver more accurate and context-aware responses and gain valuable insights into customer preferences and behavior. Not only are we aiding in the development of more user-centric products and services, NeuTigers continues to solidify itself as an innovator in the AI space.
Future Labs 370 Jay St,
7th floor Brooklyn,
New York 11201 USA
Call Us
+1 908 336 7548
Email Us
info@neutigers.com
NeuTigers USA
535 Route des Lucioles
06560 VALBONNE
FRANCE
Email Us
info@neutigers.com
NeuTigers France
Meet the Team
MHDeep
Mental Health Disorder Detection System based on
Wearable Sensors and
Artificial Neural Networks
Mental illness affects an estimated 950 million people worldwide, and is the leading cause of disability across all age groups. Stigma, lack of access to adequate mental health services and the rise in levels of stress and anxiety around us further exacerbate these facts.
In 2021, NeuTigers successfully established a proof of concept for the detection of three mental health disorders: schizoaffective disorders, major depressive disorders and bipolar disorders.
The result is MHDeep, an application that, based on inference from eight different categories of data obtained from sensors integrated in a smartwatch and smartphone, is able to deliver between 82.4 and 90.4% accuracy in detecting the three disorders.
These are the early stages of our work in this area and are devoting efforts and resources to expanding the range of applications to support patients afflicted with these as well as other ravaging conditions.
SHARKS
GRAVITAS
Cyber-Physical Systems (CPS) and Internet-of-Things (IoT) devices, often energy-constrained, struggle to implement comprehensive security measures. This limitation, compounded by the substantial data flow among devices, poses challenges in securing CPS/IoT.
NeuTigers' ML-based approach, SHARKS (Smart Hacking Approaches for Risk Scanning), systematically generates exploits in CPS/IoT. Operating at both system and user levels, SHARKS models attack behavior, explores an extensive attack space, and maps it to a defense space.
The GRAVITAS model enhances CPS/IoT device security by consolidating hardware, software, and network vulnerabilities into a single attack graph. This graph generates a probabilistic "vulnerability score," with user-defined defenses reducing vulnerability. GRAVITAS offers a comprehensive security model.
For in-depth details and performance analysis, refer to the articles on SHARKS and GRAVITAS.
DOCTOR
DOCTOR addresses the limitations of conventional machine learning methods in disease detection, which typically require individual models for each disease and lack adaptability to new data and tasks.
Our platform uses a sophisticated network and a special learning method to sequentially learn new tasks while retaining knowledge of previous ones. This unique combination enables attaching multiple heads to a single core of DNN to power simultaneous prediction, such as a single DNN simultaneously predicting multiple disease states.
It results in highly accurate yet compact DNNs, conserving space and power on energy-constrained edge devices such as wearables.