£100,000 donation accelerates prostate cancer electronic nose work
A £100,000 donation from perfume entrepreneur Jo Malone is set to speed development of an electronic nose designed to detect prostate cancer from urine samples. The funding goes to Medical Detection Dogs, a Buckinghamshire charity that has spent more than a decade working with quantum physicist Dr Andreas Mershin on a device modeled on the olfactory ability of dogs.
The project sits at the intersection of oncology, diagnostics, and machine learning. Its central premise is simple but biologically plausible: cancer alters human metabolism, and those changes can produce volatile chemical signatures in bodily fluids. Dogs can detect those scent signatures with extraordinary sensitivity. The challenge is translating that natural talent into a reproducible diagnostic platform that can work at scale in clinical settings.
The donation followed a demonstration for Queen Camilla, during which a prototype of the technology successfully detected prostate cancer from a urine sample.
Dogs, urine samples, and the chemistry of disease
Medical Detection Dogs argues that every disease has an odor and that its trained animals can smell the biochemical shifts that accompany illness. The charity says dogs have about 300 million sensory receptors, compared with roughly five million in humans. In practical terms, that means dogs can identify extremely low concentrations of odor molecules that escape routine human perception.
For cancer diagnostics, that sensitivity matters because tumor biology can alter the metabolite profile in urine, breath, or sweat long before symptoms prompt investigation. A detection system that reliably recognizes those patterns could eventually support earlier diagnosis, especially in diseases such as prostate cancer where screening remains imperfect and overdiagnosis is a real concern. The aim is not to replace pathology or imaging, but to add a rapid, noninvasive triage tool that can direct patients into the right diagnostic pathway sooner.
The charity’s current work uses trained dogs to sniff urine samples from patients. The prototype electronic nose is built to mimic that process using protein odorant receptors to capture scent molecules from urine and identify cancer-related patterns. Artificial intelligence then helps recognize recurring activation patterns across the receptor array.
A canine nose can do more than detect a scent. It can separate a meaningful biological signal from a noisy background of unrelated molecules.
SniffSpace aims to turn canine training into a larger dataset
The donation will help install a SniffSpace, a secure room that allows dogs to sniff more samples and generate more robust data for the electronic nose project. At present, the dogs sniff samples arranged in up to eight stands, and trainers manually record the results while moving the samples around between rounds. SniffSpace will automate the rotation process, which should increase throughput and reduce the time spent on repetitive handling.
That operational change sounds modest, but it is central to the science. Diagnostic AI systems are only as strong as the data used to train them. More samples, captured under controlled conditions, allow investigators to compare canine responses with the electronic nose and refine the patterns that best discriminate cancerous from noncancerous specimens. The charity said the prototype is currently “sniffing” 500 urine samples from patients at Milton Keynes Hospital, and the dogs previously identified which of those samples were cancerous.
The next step is comparative analysis: matching the dogs’ performance against the device and seeing whether the electronic nose can learn from the same scent signatures. In a field that often stalls at proof of concept, a structured dataset is what moves a biomarker idea toward a viable test.
Formal NHS device plans are already being discussed
Dr Mershin said RealNose.ai and Medical Detection Dogs are focused on establishing a formal partnership to create in vitro diagnostic medical devices for the NHS, first for prostate cancer screening and early diagnosis, then potentially for other cancers using the same olfactory approach. That ambition places the project in a broader diagnostic pipeline rather than a single proof-of-principle exercise.
The scientific logic is compelling. Protein odorant receptors can be used to trap scent molecules from urine, then artificial intelligence can interpret the receptor activation profile. If the pattern is stable, specific, and reproducible across patients, the method could offer a low-burden screening or triage test. If the pattern is inconsistent, the technology will struggle to move beyond the research setting.
This is why the quality of the dataset matters so much. Cancer detection tools fail when biological noise overwhelms the signal, or when sample handling introduces variation that does not reflect the disease itself. Automating sample rotation and collecting more observations should improve the statistical confidence around the model’s output, even before any clinical deployment is considered.
Where the technology could fit in cancer care
Jo Malone said her sense of smell has given her a great deal throughout her life and that she has long wanted to use the world of scent to give something back. Her donation gives the charity room to expand a platform that has already attracted attention well beyond the local demonstration.
Medical Detection Dogs chief executive Dr Claire Guest described the donation as game-changing and said the technology offers an unprecedented opportunity to decode scent signatures and move closer to a reliable electronic nose for cancer detection. That language reflects the scale of the challenge. Detecting one disease odor among many similar biochemical patterns is not a theatrical trick; it is a diagnostic engineering problem.
If the approach proves accurate, the most immediate bedside value would likely be in earlier triage. Men with suspicious symptoms or ambiguous initial testing could be identified faster, and clinicians might have a new, noninvasive way to decide who needs further evaluation. The device could also help standardize what trained dogs already do well, extending an ability that is impressive in a few animals to a system that can be reproduced in more places.
The work is still focused on prostate cancer and on comparing the device with canine performance in urine samples from Milton Keynes Hospital. No clinical performance figures were released, and no patient-level accuracy data were provided. The next meaningful questions are whether the electronic nose can consistently match the dogs, whether the scent signatures remain stable across larger sample sets, and whether the same olfactory strategy can be adapted to other cancers without losing precision.