SHAIPED: Institut Curie integrates a consortium of experts to create a "European space for AI in health”

In the era of artificial intelligence (AI), a major challenge for medical devices is access to real-world data to train, test and validate these tools. Their integration into health information systems as well as the proof of their clinical effectiveness also pose significant challenges. 

To address these issues, the Health Data Hub created the SHAIPED project in 2024, a consortium of 30 key players in digital health in Europe (health authorities, research institutions and public health organizations) including Institut Curie. Its ambition: to stimulate innovation in the field of AI-based medical devices, by contributing to the construction of the European health data Area¹ (EEDS) and by facilitating the implementation of the new European regulation on AI.

With a total budget of 4 million euros, the project is now officially launched and will focus on three main axes aimed at:

• Facilitating access to health data for AI-based medical devices,
• Validate the clinical effectiveness of these devices on a large scale,
• Integrating medical devices into European health systems.

Within each country, a " Data Access Responsible Body² (DARB)" will be designated in order to organize the collection and availability of data within the EEDS.

AI models to accelerate the detection of breast and lung cancers

Directed by the Léon Bérard Center in France, the use case dedicated to oncology aims to validate AI models³ in imaging, for the detection by CT of lung metastases and the detection of breast cancers by mammography. In this context, Institut Curie plays a key role by contributing to the constitution of a reference database in medical imaging, alongside cancer centers in the Netherlands, Italy and Spain. 

At Institut Curie, the radiologists of the Medical Imaging Department headed by Dr. Hervé Brisse will have to annotate 200 lung scans and 200 mammograms to form this database. This analysis will be supervised by Dr. Caroline Nhy, radiologist at the Paris site. The expertise of the Data Department will be mobilized for the identification and structuring of high-quality datasets, and for the training and evaluation of AI algorithms.

The Heath Data Hub and other European health data access organizations will collect images provided by participating centers across Europe in order to conduct validation tests of AI models, while ensuring their compliance with clinical, regulatory and operational standards. All of these data will be integrated into a secure common space within SHAIPED, making it possible to constitute a true European reference base in cancer imaging.