Owkin partners with Arkhn and Inria to launch €10M OncoLab to standardize health data access

Owkin is working with Arkhn, a French data startup, and Inria, the French national institute for research in digital science and technology, to launch OncoLab – a new project to standardize and improve access to healthcare data. The OncoLab project aims to make oncology data from healthcare institutions accessible to all ecosystem stakeholders for research and innovation purposes. The multi-modal data stored in the institutions will be standardized, structured, and studied in a synchronized manner.

A public-private consortium of expert centers to facilitate research and innovation in oncology

The OncoLab project aims to deploy data architectures for research and innovation in oncology at four leading institutions in the field: the Institut Curie, the Institut Bergonié, the IUCT-Oncopole and the Toulouse University Hospital. The objective is to respond to the various current challenges of health data management and accessibility, by providing a common and standardized technical platform for healthcare institutions and their partners. These data architectures will be developed by Arkhn, the project leader, and studied in a decentralized manner thanks to Owkin's expertise in data science and artificial intelligence, to preserve the confidentiality of the data and the sovereignty of the healthcare institutions. In total, the project has a budget of nearly €11 million.

‍

‍

Towards standardized data warehouses that guarantee the sovereignty of institutions to facilitate access to health data

The data architectures deployed by OncoLab will integrate all types of oncology data (reports, examinations, imaging, biology, etc.) for all types of cancers, collected from the hundreds of thousands of patients monitored by healthcare institutions. A secure platform will considerably simplify technical access to data for each center wishing to make it available for research and innovation projects, thus drastically reducing their costs and implementation times. This direct access to standardized data will open new opportunities for the healthcare institutions where data is produced and their partners. This will enable healthcare organizations to maintain full control over their patient data, including through modern technologies such as Federated Learning, which allows research projects to be conducted without data being extracted from the organization.

Training advanced Artificial Intelligence algorithms to better structure and exploit data for cancer research

The OncoLab project will improve the Natural Language Processing (NLP) methods needed to automatically analyze tens of thousands of medical documents (prescriptions, hospitalization reports, discharge letters, etc.) and extract relevant information for research. Developed in collaboration by the project partners, these methods are based on state-of-the-art Artificial Intelligence models created by Inria's ALMAnaCH project team.

Jocelyn Dachary (SVP IT & Data Solutions, Owkin), said that:

Ensuring that researchers have access to high quality patient data is a crucial step to future medical discoveries, which are at the heart of our mission. The OncoLab project here provides an excellent opportunity to address data interoperability, a fundamental issue in realizing the full potential of artificial intelligence. It will increase the power of the federated training already provided by Owkin technology, by adding easy access to structured data in a reference format.

Emeric Lemaire, Corneliu Malciu and Théo Ryffel, co-founders of Arkhn, said:

The OncoLab project is a great opportunity to give healthcare institutions access to their patient data, while paving the way for research projects and public-private partnerships that benefit the entire ecosystem.

Nicolas Portolan (Deputy Director of Bergonié), said that:

Inventing a common language, thanks to the contributions of AI, will make it possible to remove many of the obstacles that healthcare players face today and make it much easier to answer the questions of our clinicians/researchers. An Esperanto algorithm to accelerate cancer research!

Julien-Aymeric Simonnet (Director of Data and Digital Health), said that:

The mission of the Oncolab project consortium is to make heterogeneous and massive medical and biological data speak for themselves by structuring them. This is a fundamental building block for conducting research using real-life data and developing medical innovations for the benefit of patients.

Eric Fleury (Director of the Inria research center in Paris), said that:

The applications of digital sciences in the health sector are at the heart of Inria's strategic axes. The Oncolab project is a perfect illustration of the contribution of AI and the diversity of public and private partnerships in the service of health.

Jean-François Lefebvre (Director of the Toulouse University Hospital), said that:

"The Oncolab project is the result of a fruitful collaboration between the Toulouse University Hospital and the Claudius Regaud Institute, both of which are part of the Toulouse University Cancer Institute - Oncopole, with the company Arkhn. This strong link between the Toulouse University Hospital and the startup ecosystem aims to structure cancer health data on an unprecedented scale. The integration and development of AI technologies in our data warehouses opens the doors to data research by facilitating their reuse to improve knowledge of the disease, always for the benefit of patients. This is a large-scale project, which reflects the excellence of the Toulouse University Hospital teams, who are fully invested in digital innovation."

Pr Jean-Pierre Delord (Director of the Claudius Regaud Institute and administrator of the GCS IUCT-Oncopole), said that:

OncoLab fits perfectly with our AI partnership strategy. Data structuring is a major challenge to promote the care-research continuum and is one of the strengths of the IUCT-Oncopole.

Amaury Martin (Deputy Director of Institut Curie), said that:

The Oncolab project is a great opportunity for the Institut Curie to share its expertise in data standardization, which was initiated several years ago, within a leading consortium, and to amplify this dynamic at the national level before looking to the international level. Data-driven cancer research will necessarily involve dialogue between decentralized databases, placing access to harmonized technical bases and sharing standards at the heart of the issues.

‍

About Arkhn

Arkhn responds to the issue of accessibility of health data by mobilizing its unique know-how in the development of standard health data warehouses within health institutions. This approach allows healthcare institutions to maintain their sovereignty over the data they produce, while promoting the development of a research and innovation ecosystem around the use of this data.

‍

About Inria

Inria is the French national institute for research in digital science and technology. World-class research, technological innovation and entrepreneurial risk are its DNA. Within 200 project-teams, most of which are shared with major research universities, more than 3,500 researchers and engineers explore new avenues, often in interdisciplinary ways and in collaboration with industrial partners, to meet ambitious challenges. As a technology institute, Inria supports the diversity of innovation paths: from open-source software publishing to the creation of technology startups (Deeptech).

The ALMAnaCH (Automatic Language Modelling and Analysis & Computational Humanities) project-team is part of the Inria research center in Paris. ALMAnaCH's research field is automatic language processing (ALP), at the heart of Artificial Intelligence and Digital Humanities, at the crossroads between theoretical computer science, machine learning and linguistics.