MIRA – Medical Innovation for rare disease and medical imaging using AI

The project aims to shorten the diagnostic journey for rare diseases through the application of advanced digital technologies and artificial intelligence tools.

PROJECT LINE A
AI-supported diagnosis of rare diseases

The Project Line A is dedicated to developing solutions for the suspicion and identification of rare diseases based on the analysis of textual clinical reports, with the goal of unlocking the value of information that is often available only in unstructured form.

OBJECTIVES

The initiative seeks to transform textual data contained in clinical documentation into shareable, accessible, and interoperable information, so that it can be used by artificial intelligence algorithms and new digital tools supporting diagnosis. The project therefore addresses one of the sector’s main challenges, namely the difficulty of extracting, organizing, and integrating heterogeneous data originating from different clinical contexts.

To achieve this goal, the project includes the development of tools for the automatic extraction of data from clinical reports, the definition of standard models for their structuring, and the creation of a digital ecosystem for the secure management, sharing, and analysis of information. The adopted approach follows FAIR-by-design principles, ensuring that data are findable, accessible, interoperable, and reusable, in compliance with regulatory, ethical, and privacy protection requirements.

A central part of the activities also concerns the development of virtual assistants based on Natural Language Processing techniques, capable of automatically recognizing clinically relevant entities and concepts in textual reports and converting them into standard terminologies. These tools are designed to provide researchers and clinicians with new support in organizing and interpreting information, helping to make the analysis of clinical data in the diagnostic field more effective.

The project also provides advanced computational infrastructures for the training and execution of machine learning and deep learning models, in order to support the processing of large volumes of data and accelerate the development of artificial intelligence applications in healthcare. Overall, the initiative aims to build an intelligent platform for the diagnosis of human diseases, using rare diseases as a reference model for developing innovative, scalable, and transferable tools that can also be applied to other clinical contexts.

RESULTS

The project aims to develop automated tools for the extraction and standardization of textual clinical data, define a shared format for their representation, and build a platform for the secure and interoperable management of information. The objectives also include data protection and governance, the development of artificial intelligence–based virtual assistants for report analysis, and the provision of high-performance computing infrastructures to support analytical activities.

The expected outcome is the creation of new digital tools capable of facilitating the identification of patients with rare diseases on the basis of specific characteristics detected by artificial intelligence, thereby contributing to a faster, more accurate, and more structured path toward diagnosis.

PROJECT LINE B
Automated evaluation of diagnostic images through artificial intelligence 

The project is dedicated to the development of artificial intelligence solutions for the automated analysis of diagnostic images, with the aim of supporting the recognition of anomalies or lesions and contributing to the identification of rare diseases. Project Line B focuses in particular on the ability of algorithms to process and classify different types of radiological images, providing advanced tools to support diagnostic activities.

OBJECTIVES

The initiative stems from the need to enhance the value of data generated in clinical practice, making diagnostic images more easily usable in automated analysis and decision-support pathways. The project therefore includes the comparison of images from healthy patients, patients with rare diseases, and patients affected by other conditions, in order to develop artificial intelligence tools capable of identifying distinctive features useful for recognizing specific clinical conditions.

A significant part of the activities concerns the definition of a standard format for the extraction and structuring of data associated with diagnostic images, with the goal of promoting interoperability across different sources and creating the conditions for the effective use of analysis algorithms. At the same time, the project includes the development of machine learning and deep learning systems for the analysis of radiological images, such as MRI scans, CT scans, PET scans, and X-rays, including in contexts where the availability of labeled data is limited.

The project also includes the creation of an infrastructure dedicated to supercomputing and artificial intelligence, designed to support the training and execution of complex algorithms on large volumes of data. Overall, the initiative aims to provide new digital tools for the automated evaluation of diagnostic images and for supporting the diagnosis of rare diseases.

RESULTS

The project aims to develop artificial intelligence tools for the automated analysis of diagnostic images, define a standard format for the extraction and organization of data, and build a technological infrastructure suitable for the training and execution of advanced algorithms. The expected outcome is the availability of digital tools capable of improving the reading and evaluation of diagnostic images, thereby helping to make the diagnostic pathway more effective.