2022 – 2028
ANR funding: Priority research program on Rare Diseases
CDE.AI: Artificial Intelligence at the service of common data elements for rare diseases
The BNDMR (Base de Données Maladies Rares) evaluates the management of rare diseases, by collecting Common Data Elements. These elements are collected manually. Our aim is to create CDE.ai, a set of natural language processing (NLP) algorithms that automatically fill in the CDM collection forms, directly from patient records.
Medical records (10,000 patients) will be used to train and evaluate CDE.ai. A broad spectrum of diseases will be covered, mainly rare autoimmune and auto-inflammatory diseases – with patients treated with glucocorticoids – from two rare disease networks (FAI2R, and FILNEMUS), and rare endocrine diseases – with patients with endogenous glucocorticoid excess – from a third network (FIRENDO). For all these patients, CDE.ai will be used to collect the CDE for the BNDMR. This BNDMR register will be used to select patients with glucocorticoid excess (2,000 patients). For these patients, CDE.ai will be used to collect the CDE for a database of glucocorticoid complications. A biobank will be set up (blood and urine, 300 patients), to measure molecular markers known to be associated with glucocorticoid sensitivity, in search of predictive markers.