OWKIN: The Da Vinci Mode(l)

Leonardo Da Vinci didn’t just paint the Mona Lisa and The Last Supper. He also designed the helicopter, army tank, studied anatomy, and devised a system of movable barricades to protect the city of Venice from invasion. Today, he would probably have invented KickStarter and become CEO of a Bcorp. In short: he excelled in many different areas, instead of specializing in just one.

His different interests and varied skills informed and strengthened each other - his ability to sketch anatomy would undoubtedly have helped him draw the ‘anatomy’ of an experimental piece of engineering. This is called cross-fertilization. And this is the key to innovation according to ....me.

This is the kind of overarching, orthogonal approach that we have built at Owkin . Call it, ‘The Da Vinci Model’.

Whereas many other companies in our space take a vertical approach - specializing in data and analytics, drug discovery or drug trial optimization - Owkin is building the first fullstack AI Biotech for precision medicine, by design. From patients, to patients, and for patients.

It works like this:

1. Our fullstack starts with multimodal patient data and samples from tier 1 academic centers- the backbone of our business. By building relationships with the most advanced research hospitals we are able to find the right data to answer the most pressing challenges of our pharma partners, along with our internal research efforts.?
2. The second layer is about integrating AI-derived biological insights with wet lab data from experiments in what we call “lab- in-the-loop” - validating early assumptions and feeding experimental data back into our AI-engines in a powerful feedback loop that accelerates R&D.
3. The final element is, of course, artificial intelligence. We are creating multiscale causal AI to connect the dots across cellular, molecular, and tissue scales, capturing the causal links of complex biology. This data-driven approach reveals biologically distinct subgroups of patients that form the basis of a novel, multiscale understanding of disease

Sample, isn't it?

We believe our unique data access and AI will enable us to double the probability of success across drug discovery and development.

But while discovering and developing more and better drugs is at the heart of our mission, we are also investing in developing cutting-edge AI diagnostics that will help match the right patient with the most appropriate treatment.?Our vision is a world of research and care, connected. Where medical experts, augmented by machine learning and large-language model tools, identify the most pressing questions - and our causal AI, powered by patient data, can answer them.

?But to achieve this requires powerful AI, the right data, and a deep understanding of biology. The way we tackle these challenges sets Owkin apart:

Unparalleled access to multimodal patient data

How can one better understand and portray people in all their complexity? This question drove much of the advances in art and science during the Renaissance. In the present day, it’s possible to amass vast quantities of patient data, but to still poorly understand the patients behind the data.

?Instead, at Owkin, we build our understanding by collaborating with academic PIs across the world - including 120 top clinical researchers at top hospitals in the US and Europe. This allows us to access up-to-date multimodal longitudinal data and samples, that captures the ever-evolving standard of care, to train our models. Owkin is also investing in partnerships like MOSAIC to generate rich modalities like spatial omics that give us a fuller, heterogeneous picture of patients and their disease.

?A key driver of our ability to scale our data network globally is our focus on privacy enhancing technologies, like federated learning that give our academic partners confidence to work with us. This privacy-respecting model has allowed us to collaborate with 8 of top 20 research hospitals in the US and Europe.

Cutting-edge AI

?Medieval art was flat and two dimensional. In two words, boring. Renaissance artists, like Da Vinci, changed that by bringing depth, scale and perspective into their portraits. In two words, less boring. Owkin’s AI integrates data from across different modalities and over different biological scales, from gene to tissues, unlike many of our competitors, to bring depth and scale to our models.

?At the heart of our AI is an understanding of causation, rather than correlation. Our cutting-edge algorithmic research, and creation of foundational models, allows our models to better capture how our potential targets are causally affecting the diseases we work in, and how we understand links between different scales, from genes to cells, cells to tissues, and tissues to diseases.

This increases our confidence in their suitability for preclinical development. By also integrating prior knowledge, through our large-language models, we further strengthen that confidence. Compare this to other systems, still using only correlational AI, and they will miss these causal connections and their drugs will fail to make it to patients.

?But our AI always works in concert with our human experts - this is the Owkin Cortex: the best of human and artificial intelligence.

To make this work, we have ditched the black-box approach of other players in this space: our AI is interpretable. Being able to look under the hood lets us understand better why our algorithms make the decisions they do and therefore continually improve the model performance.

Grounded in biology

?500 years ago, people wanted to discover new biology and become the Elon Musk of their time. This desire hasn’t changed, people still want to become Elon Musk. Unlike other companies in this space, who focus simply on the output of their AI, our goal is to uncover the biology underlying the disease - for example to understand tumor microenvironment interactions at the cellular level.

This fuller understanding of biology allows us to categorize patients into subtypes based on their differing biology - the idea that underpins all of our R&D work. That in turn allows us to identify specific biomarkers for those subtypes that can drive our target discovery engines. Once our AI has suggested targets, we can then validate them in our wet labs - that experimental data can then be fed back into our AI models to refine our search for new targets.

?Da Vinci famously said that “the noblest pleasure is the joy of understanding”. Probably before he arrived to his "incubator" at Versailles, the YC of the time with the best venture capitalist of all times: Francois the 1st.

At Owkin, we are driven by a strong desire to understand complex biology through AI and to use that understanding to bring better treatments to patients sooner.

We believe that by tackling the combined challenges of drug discovery, development, and diagnosis, not just part of it, we can achieve more than others could even imagine.