reimagining the relationship between humans and our best friends
- Dogs can do what humans cannot, such as detecting 28+ diseases in real-time, often with higher accuracies than our best diagnostics.
- Today, there are more dogs on Earth than ever in human history. The United States had more canines than children in 2020 and data from Brazil, China and India point towards population booms worldwide.
- For the first time ever, we have the ability to map brain activity using non-invasive, highly portable, and relatively inexpensive devices.
Leveraging non-invasive brain-computer interfaces powered by EEG/fNIRS and powerful AI algorithms, we can begin to decipher the canine mind in real time.
This has three major potential cascades —
- Symbiosis - Increasing the bandwidth of human-canine communication, cooperation and welfare.
- Partners not Pets - Increasing affordances available for all dogs to be free-agents capable of providing for each other and our collective health, security and understanding.
- Unleashing Disease-Sniffing Dogs - Real-world applications of canine olfaction to allow for cost-effective, accurate, and non-invasive tests.
With modern and comfortable custom-designed canine BCI systems facilitating high bandwidth data, the ability to classify and predict perceptual stimuli, emotional states, coarse cognitive thoughts, personality traits and brain health/functioning is scientifically sound.
Recent studies show that scalp EEG can provide meaningful inferences on face processing (Nature Scientific Reports, 2020), human vs dog vocalizations (Royal Society, 2022) and known vs unknown words (Royal Society, 2020). These inferences were possible despite limitations from using wired human systems and electrodes that are several years behind current state-of-the-art as well as relying on sparse data (few dogs and total minutes recorded).
Papers -
Towards A 4E Approach to Canine Olfaction (link)
Non-Invasive Canine Electroencephalography (EEG) : A Systematic Review [preprint]
Towards Canine BCIs [preprint]
A Human-Canine Interface from Sensors, Haptics, and AI [preprint]
Learning about disease and cognition in dogs can have profound ramifications for humans.
A growing consensus is emerging that dogs are a highly robust model for comparative neuroscience and medicine. This is possibly because of -
- Our shared co-evolution and living environment.
- Our shared propensity to naturally develop conditions such as dementia, epilepsy, and ADHD, among others.
- The natural experiment emergent from the tremendous neural and genetic variation amongst breeds.
This is why companies like Rejuvenate Bio (George Church) and research efforts like the NIH-funded Dog Aging Project are focusing their longevity efforts on dogs, with Loyal being a pioneer having received FDA approval for their longevity treatment for large dogs.
Detection dogs has been evaluated in detection multiple types of diseases and biological conditions in humans, animals, and even plants. A handy table from a 2021 paper summarizes twenty papers
The above table does not include the large number of studies investigating detection dogs for Covid19, summarized below in a table from the same paper -
For a brief discussion, see the Substack essay Canine Medical Intuitives. One point worth highlighting is that, in one study that explicitly compared the two, detection dogs were more accurate than gold standard PCR for Covid-19 (Nature, 2021).
Despite increased interest in devising machine olfaction approaches to disease and other kinds of detection, dogs appear to retain the best combination of accuracy, practicality, reliability and cost-effectiveness, for olfactory detection.
A robust platform to collect and understand canine neural activity is technologically feasible.
Tremendous progress is being made in skin-interfaced bioelectronics, e.g. liquid metal-elastomer composites (Science, 2023), and the future of EEG is rift with innovation (Wearable EEG and Beyond, 2019). At the same time, progress in artificial intelligence is at an inflexion point and AI systems are already being used to make sense of neural data, with some success (Stable Diffusion can reconstruct images from fMRI data). The field of machine learning models in EEG analysis continues to develop, with deep learning maturing in real-world settings (IEEE, 2022).
A growing number of human EEG studies use commercial consumer products such as Muse, with clever algorithms capable of by-passing constraints imposed by noisy data from a limited number of electrodes. There is no reason the same cannot apply for dogs. While difficult challenges certainly exist, such as deriving high-quality SNR during motion and through fur and a thicker skull, and designing suitable form factors for the large variation of canine skulls, they are not impossible to surmount and imperfect MVPs will still be able to deliver considerable use-cases whilst providing data to further refine models.
Dognosis resonates with the trifecta of poiesis technology, dialogical living and nosing towards aligned uplift, articulated in a manifesto concentrate —
"Relations are constitutive; dogs and people are emergent as historical beings, as subjects and objects to each other, precisely through the verbs of their relating." Donna Haraway, When Species Meet
In motion -
- Deploying the most functional EEG headset in the world to collect experimental-grade canine EEG data
- Applying signal-boosting algorithms and developing ML predictive models for canine EEG data
- Working with collaborators (Sympoiesis Technologies) on a novel human-canine interface catalysed at Zuzalu.
- Launch of the largest breath-based multi-cancer screening study in the world in Bangalore, India.
Past -
Emergent Ventures India
Boost VC
1517 Fund
gradCapital
Govt of India and Telangana State T-Hub health-tech program
Dognosis is an ambitious, long-term project whose success entails fundamentally new paradigms of understanding the world — how we relate with other species, the way we conceptualize the role of technology, and the manner in which we practice medicine.
We are looking for curious and passionate people to join us on our mission. More about who we’re looking for here. In short, we are looking for people to solve novel and challenging problems across domains of neurotechnology, electronic/mechanical design, olfactory science, embedded systems, signal processing, computational modelling, cognitive neuroscience and machine learning.
LinkedIn - https://www.linkedin.com/company/dognosis
Instagram - https://www.instagram.com/dog.nosis