UC Berkeley grad behind a multi-cancer screening tool harnessing a dog's superpower: its sense of smell

BERKELEY, Calif. - A UC Berkeley graduate is behind a new early detection, multi-cancer screening tool that pairs the incredible scent-tracking ability of dogs with artificial intelligence. 

What we know:

Akash Kulgod co-founded Dognosis, a start-up which is seeking to create a highly accurate diagnostic test that’s affordable, simple, non-invasive, and widely accessible.

Headquartered in Bangalore, India, the research and development biotech firm uses dogs that are trained to sniff and detect volatile organic compounds (VOC) that are associated with cancer. 

The dogs are then a part of a tri-layered bio-AI technology system created to digitize canine olfactory responses to "read dogs at a superhuman level."

How it works

The result is a product called BreathEasy. Designed to be simple for users, it detects cancer risk from a single exhaled breath sample, according to the company. 

To get a screening, a person breathes naturally into a custom mask for 10 minutes, covering the nose and mouth, to allow the mask to capture volatile organic compounds from the breath. 

The user then carefully seals the mask in the special packaging, placing it in a pre-labeled kit that's provided, and ships it out to Dognosis’ diagnostic facility.

Once the sample is received, the mask undergoes analysis, in which trained detection dogs sniff the sample. Dognosis uses its canine-AI system to generate a volatile organic compounds (VOC) health score. 

Clinical Trials

Dognosis has been running clinical studies in India, at partner hospitals and clinics, as well as community screening camps, in its effort to detect cancer at the most treatable early stage.

The start-up said, to date, it has conducted the largest multi-cancer breath-based detection study.

By the numbers:

In a peer reviewed study that was published in the Journal of Clinical Oncology in April, Dognosis researchers evaluated a cohort of 1,502 participants and found its breath-based cancer detection product correctly found cancer in 90.8% across seven cancer types and correctly detected no cancer in 91.3%.

The seven cancer types were breast, cervical, colorectal, lung, oral, ovarian, and prostate.   

UC Berkeley grad

The backstory:

Dognosis' co-founder Akash Kulgod graduated from UC Berkeley in 2022 with highest honors in Cognitive Science.

Kulgod said his honors thesis, which earned him the department’s prestigious Glushko Award, established "the scientific foundation" for his start-up.

He founded Dognosis in 2024, in his home country of India with co-founder Itamar Bitan, a former member of the Israeli Defense Forces’ elite K9 unit and former head of operations of an Israeli startup focused on disease-detection dogs.

VC funding

Local perspective:

The start-up has completed two rounds of funding, with Bay Area-based investors among its backers. 

They include The House Fund, a venture capital firm that focuses on investing in the UC Berkeley community, Stanford-based Arben Ventures, which invests in deep-tech health start-ups, as well as Boost VC and Draper Associates, both based in San Mateo.

Dognosis raised $2 million in its seed round. The second round of funding, led by VC firm Accel India, was held in March. Dognosis has not yet disclosed how much was raised then. 

Why India?

Kulgod, who comes from a family of doctors in the southwestern Indian city of Belgaum, said the ability to run large clinical trials at a significantly lower cost at a much faster pace were some of the reasons why they chose to start up Dognosis in India. 

"We also needed a large and interdisciplinary team and the talent of India is top-notch," Kulgod explained to KTVU in an email correspondence.

As the deep tech firm looks to expand, Kulgod said the company is working to establish a U.S. headquarters.  

Using dogs for disease detection 

Dig deeper:

Using dogs to sniff out disease is not a new concept. A 1989 publication which appeared in the medical journal Lancet, is widely considered the first documented report of a dog that was able to detect cancer. 

The publication "Sniffer dogs in the melanoma clinic?" described a dog that was fixated on a mole on its owner's leg. The growth ended up being malignant melanoma.  

Since then, there has been a wide breadth of research on canine olfaction as a way to detect cancer and other diseases.

And in this age of AI, Dognosis is using the technology to take it a step further, in its quest to create a more consistent, more accurate mode of detection, to remove human interpretation out of the equation. 

SEE ALSO: Robot house cleaning service is being offered in this Bay Area city for $30 an hour

"The problem has always been scaling that into something medicine can standardise and trust. Most detection dog work trains a specific alert, a sit or a bark, compressing a dog's rich, natural response into a single decision, and a handler then has to read that alert live, a judgement call that varies from person to person and day to day," the company wrote in a frequently asked questions document. 

What Dognosis is trying to do is to use machine learning to create that standardized reading. 

"So the result isn't dependent on one dog or trainer's read on one day. That pairing of biological detection with computational analysis is what lets us build something sensitive and repeatable at scale, rather than a one off demonstration," explained Manal Shakeel, communications lead at Dognosis. 

3-pronged technology stack

To create that computational analysis, Dognosis created a tri-layered system allowing it to build a screening tool at a real scale.  

SniffSpace is a fully automated olfactory workstation, where dogs are presented with each sample to sniff, with their reactions being captured as they occur. 

Each dog is outfitted with what's called DogSense, described as a "brain-computer interface" which provides data that a camera can't see, including the dog’s neural activity, heart rate, and breathing.

DogOS is a machine learning system, designed to decode all the data together, "teaching machines to read dogs at a superhuman level," the company said.

Based on the patterns identified, the system generates a VOC health score.  

The information is used to write up a report which is then shared with the person’s clinician, "giving them the information they need to recommend appropriate next steps for your health," Dognosis explained.

'Triage tool' 

It’s important to note that Dognosis’ product is designed to serve as a triage tool rather than a standalone diagnostic for a broad population.  

It’s meant to flag disease so a person who is shown to have cancer can seek further tests that are already the current standard of care, according to the start-up.

"BreathEasy is built to sit at the top of a cascading funnel rather than solve this alone. Its job is to narrow a broad population down to smaller, higher-risk group, cheaply and non-invasively," the company explained. 

How its dogs are treated 

The company said it has a deep sense of care and connection with the dogs its team members train.

And its commitment to its canine partners goes beyond the lab. 

When one of its dogs named Jessie was forced to retire early because of health issues, she ended up being adopted by a company team member who knew the dog and wanted to give her a home.

"That's the model we lean on - a dog from our pack doesn't just leave once its diagnosis days are behind it, it stays with us, rather than going somewhere unfamiliar," the company explained, adding, "Care doesn't end once the sniffing does."

Kulgod himself is a dog lover and has been surrounded by dogs his whole life, saying he grew up with "a rotating cast of dogs." 

What type of dogs 

The company also said its dogs are from both shelters and ethical breeders in the Bangalore area.

While the majority of its pack is beagles, researchers do not seek specific breeds. 

What matters is temperament: a dog that trusts the handler, genuinely loves to sniff and play, and loves food.

Beagles happen to fit the description pretty well, so the breed makes up most of the pack 15 dogs it’s currently working with, but there are other kinds of dogs that also match that temperament. 

"One of our best-performing dogs, Chloe, is a shelter mix between an indie street dog and a Labrador, and she holds her own against our beagles," the company said. 

What's next:

The start-up said it plans to launch in India early next year, allowing doctors to recommend BreathEasy as part of standard patient care.

Also next year, Dognosis plans to open a U.S. headquarters and said it's currently searching for ideal sites to set up a canine training facility.

Kulgod told KTVU the Bay Area is one of the locations the company has in mind, along with Texas.

Ultimately, Dognosis said its goal is to scale its technology worldwide with the screening tool being easily and widely accessible.

"Our ambition is for BreathEasy to become a standard part of people's annual health checkup, not a test people seek out only when something feels wrong," the company explained. 

The company acknowledged all the partners involved in its quest to make that a reality. 

"Our work increases the bandwidth of human-canine communication and collaboration," the website states, noting the company was formed by "a group of ambitious humans and dogs working towards reimagining the possibilities of interspecies partnership."