A collaborative project combining the use of health data and artificial intelligence, Deep.piste was selected as a pilot project by the Health Data Hub (HDH). The announcement was made on April 16th by Bernard Nordlinger, President of the Selection Committee, during a ceremony attended by French Minister for Health and Solidarity Agnès Buzyn.
Deep.piste is led by Epiconcept’s Data Science unit manager Francisco Orchard, in partnership with the Regional Cancer Screening Coordination Centre of the Occitanie Region and Professor Fabien Reyal’s research team at the Institut Curie. Integrated into e-SIS, Epiconcept’s cancer screening campaigns management software, the project offers an implementation of AI for 2nd or 3rd mammography readings.
The dedicated analysis algorithm will be developed under an open license. The HDH jury declared it was “impressed” by this desire to “share data without reservation” with the scientific and medical community. Acknowledging the project’s public interest & innovative use of health data, it will support Deep.piste for one year, providing for shared human and technological resources.
Deep.piste in detail:
With an estimated 11883 deaths in 2017, breast cancer is the most deadly disease for women in France. Early detection makes it possible to reduce mortality by 21%. Deep.piste proposes to use cross-referenced sources to evaluate the impact of AI intake in the organized breast cancer screening system:
The e-SIS base of anonymous breast cancer screening data from Gard and Lozère French departments (more than 250,000 images annotated in ACR classification);
Data from the French National System for Health Data that could help identify significant medical events outside the screening circuit (causes of death, reimbursement data, etc.), in order to reduce the number of “false negative” results.
Exploited in a scenario-based logic and in support of dense convolutional neural networks (deep learning), they will participate in the screening process in a new way.
The advantages of the project for Epiconcept:
On the Epiconcept side, the collaboration with the Health Data Hub is a first opportunity to make public health data available and use it. Key skills should ultimately be offered to the company’s healthcare clients (research teams, epidemiologists, etc.), such as a mastery of administrative and industrial procedures for data access, or a better understanding of their added value.
The project’s assets for the CRDC-OC:
Since 2014, the Occitanie region, with the departments of Gard and Lozère, has been conducting an innovative and avant-garde experiment in the dematerialization of breast cancer screening stream of images. The project could participate in the improvement of cancer screening.
Today, 60% of French departments trust Epiconcept to support the daily management of cancer screening campaigns. e-SIS, its complete and 100% web-based management solution, is designed to support:Organized breast cancer screening;
Organized cervical cancer screening;
Organized colorectal cancer screening.
Common to management centres and practitioners, e-SIS makes it possible to optimize the patient file process.