Fueled by new technologies like blockchain and artificial intelligence (AI), the medical industry is innovating rapidly. One of the companies standing out in this arena is Owkin, an AI-based startup that seeks to facilitate medical research collaboration.
Recently, the startup secured $18 million from Mubadala Capital and Bpifrance through a Series A extension. Altogether, Owkin has raised $70 million in capital for its machine learning platform that connects academic researchers, clinicians, data scientists, and pharmaceutical companies.
The platform makes it possible to generate global datasets between multiple stakeholders, enabling worldwide collaboration. Owkin is dedicated to breaking down information silos and making it easier for research to occur, thereby bringing better health solutions to patients at a faster rate.
Dr. Thomas Clozel and Dr. Giles Wainrib joined forces to cofound Owkin in 2016. Dr. Clozel is a clinical research doctor with expertise in hematology and Dr. Wainrib has worked to develop the application of AI in biology. These two founded the company on the belief that medical research can and should be more collaborative and inclusive while maintaining patient privacy.
Here’s what you need to know:
Understanding the Owkin Platform for Medical Researchers
The foundation of the Owkin platform for connecting medical researchers is known as the Loop. This Loop allows researchers from top academic centers around the world to share their high-quality datasets and gain access to information collected by other scientists.
The datasets available through the Loop are curated by leading experts who point out gaps in data and identify potential biases to improve the outcomes of clinical research. The database currently includes molecular, imaging, outcome, and pathological data. The Loop is powered by the Owkin Software Stack, which is made up of Owkin Studio and Owkin Connect.
The Studio is a software platform that medical researchers can use to apply AI to their own scientific questions and research information. The company has worked diligently to make the Studio accessible and user-friendly so that researchers do not need to have expertise in data engineering to use it.
The models that are developed through the Studio help researchers pinpoint the specific data points responsible for unexpected predications so that they can narrow their research questions appropriately. Owkin intends the Studio to be used for discovering new biomarkers that can help make medicine more tailored and predictable.
The other half of the Software Stack is Owkin Connect. This software enables Federated Learning through the Loop. Data scientists are able to create AI models using data that has been decentralized to make use of information collected by other labs.
Each medical center is encouraged to use a firewall that keeps its data secure and private while allowing other users to access the information for the purposes of scientific discovery. Each data owner can define authorizations and track usages in a blockchain-like system that creates an immutable ledger. The Federated Learning model keeps data quality high while allowing research to comply with data privacy regulations.
Driving Research with the In-House Owkin Lab Algorithms
The Owkin research platform also includes the Owkin Lab. This lab features an in-house team of award-winning scientists with expertise in data, biology, and statistics. These individuals collaborate with the lab’s partners to train them on AI models and help design new research solutions tailored to the questions that will drive their future investigations.
The Lab currently features a number of Kaggle grandmasters and a DREAM Challenge top performer, as well as 14 PhDs. Furthermore, the lab is divided into specific teams, each with its own expertise on data modeling. These teams’ focus areas include clinical data, pathology, radiology, genomics, and clinical data. Already, Owkin Lab and its partners have published in top scientific journals like Nature Medicine.
Owkin has already created a catalogue of 30 live models and has more in the development pipeline. The models were all constructed with Loop or public data and are now available through Owkin Studio for use by medical researchers, as well as pharmaceutical partners through Owkin Solutions.
The company has developed an expertise in hepatocellular carcinoma, the most common form of primary liver cancer in the world. Researchers can now access drug development models for the cancer that have been acquired through machine learning and designed to increase the efficiency of clinical trials. The unique structure of Owkin makes it possible to develop such expertise in other key diseases in the future to facilitate research.
The models developed by Owkin fall into three primary categories. The first involves data enrichments. These models are for pathologists and translational researchers. Through the models, researchers can more effectively identify molecular markers through histological studies. The data enrichment models primarily facilitate biomarker identification.
The second category involves outcome prediction and help drug developers design better clinical trials and ultimately evaluate them more efficiently. The outcome prediction models deal primarily with hematology and oncology. However, Owkin has also designed some for autoimmune diseases like lupus.
The third category of models help identify patients that might benefit from a specific treatment. These models are meant to drive development in the field of precision medicine.
