CAMBRIDGE, Mass., Aug. 31,
2023 /PRNewswire/ -- Osmo, a pioneer in AI and scent,
has published a seminal paper in Science, entitled "A
Principal Odor Map Unifies Diverse Tasks in Human Olfactory
Perception," in collaboration with the Monell Chemical Senses
Center, University of Reading, and Arizona
State University. Spun out of machine learning research done
at Google Research, Osmo brings AI and olfactory science together
to give computers a sense of smell.
Two questions have long intrigued olfactory researchers: why do
molecules have specific smells, and how can we predict those smells
based on their structures? Over the past century, sparse data, lack
of advanced technology, and the perception of smell as a
qualitative, subjective experience have limited progress in
answering them. But with recent advances in AI, Osmo has been able
to solve these long-standing challenges—marking a pivotal moment
for engineering olfaction.
Using graph neural networks (GNN), Osmo CEO Alex Wiltschko and his team created a principal
odor map (POM), the first generalized odor map of its kind that can
outperform people at scent prediction. Their machine-learning
model, which was trained on a dataset of 5,000 molecules, predicted
the scent of hundreds of molecules that had never been smelled
before, based solely on molecular structure. Comparing the model's
performance to that of individual participants, who rated scents
using a lexicon of odor descriptors, it achieved better predictions
of the consensus—the average of the group's odor ratings—than any
single participant in the study. It was also able to identify
dozens of pairs of structurally dissimilar molecules that had
counter-intuitively similar smells, and to characterize a wide
variety of odor properties, such as odor strength and perceptual
similarity, for 500,000 potential scent molecules.
"Our work represents the very first step to quantifying our
sense of smell," said Alex. "Computers have been able to digitize
vision and hearing, but not smell—our deepest and oldest sense. The
fundamental nature of smell is that it serves as the basis of human
survival and plays a formative role in our emotions and memories,
yet we haven't been able to develop systematic methods to quantify
this important sense. Now that we can control and engineer scents,
we will finally be able to take the next step of innovating
olfaction, in order to benefit human health and well-being."
Based on this research, Osmo has created a powerful platform
that accelerates the search through the billions of molecules in
chemical space to predict the scent of any molecule that exists or
is manufactured.
"This is a zero-to-one moment for olfaction," said Alex. "AI has
transformed so many processes in the chemical world, including drug
discovery, but scent hasn't been tackled yet. In some ways, we see
AI in drug discovery as a precedent for ways that AI will
revolutionize olfaction. Similar to how the advent of AI is
changing the way drug hunters find new candidates that are more
likely to succeed in the clinic, we see AI augmenting the role of
synthetic chemists and master perfumers, the scent hunters who comb
through myriad molecules to create the scents that are an integral
part of our lives."
Moving forward, Osmo is pursuing two main goals to support its
mission. The fragrance industry is facing stricter standards and
more difficult design challenges, so the company's first goal is to
lead the discovery of new scent ingredients that are safe,
sustainable, and environmentally friendly. The second goal is to
digitize smell. In the same way that RGB and frequency mapping
opened up a world of technological innovation for vision and
hearing, Osmo aims to create digital representations of smell that
will change how we capture, transmit, and remember scents. This
reflects the underlying vision of Osmo—to catalyze the evolution of
scent from ephemeral to enduring.
About Osmo
Launched in January
2023 with $60 million Series A
funding led by Lux Capital and Google Ventures, Osmo fuses machine
learning, data science, psychophysics, olfactory neuroscience,
electrical engineering, and chemistry in a multi-disciplinary
approach to digitizing scent. Osmo's work is grounded in digital
olfaction research that the team validated at Google Research,
including two pivotal studies that used Graph Neural Networks to
predict the smell of a molecule from its structure and to
investigate the biological underpinnings of odor similarity. The
company has begun work in the flavor and fragrance market to create
a new generation of better, safer, environmentally-friendly aroma
molecules. Over time, Osmo expects to work in domains such as
public health and agriculture on solutions that help humans detect
diseases earlier, track pandemics faster, grow more food, catch
food spoilage, and ward off insects.
Media Contact
Jeannie
Entin
jeannie@osmo.ai
View original content to download
multimedia:https://www.prnewswire.com/news-releases/digital-olfaction-startup-osmo-introduces-scent-to-the-world-of-artificial-intelligence-in-breakthrough-science-paper-301915197.html
SOURCE Osmo