How can AI be used to stratify the risk of rare hereditary diseases? (CardI-HACK project)

Hypertrophic cardiomyopathy (HCM) is an inherited heart disease that is a major cause of sudden death and/or heart failure in young people <40).

• This relatively rare disease affects 1 in 500 people in the general population.
• It is defined by an increase in the thickness of the left ventricular paroid of the heart that is not explained by load conditions.
• Until now considered monogenic, researchers and doctors have long thought that this disease was linked to a mutation affecting a single gene in the patient.
• Despite improved knowledge and management of hypertrophic cardiomyopathy, the disease remains unexplained in more than 1 in 3 cases.

Due to its rarity, the various clinical and genetic data collected for this disease are heterogeneous, and often based on separate patient cohorts.

At present, cardiogeneticists at rare disease centers are all too often at a loss when faced with the results of genetic and cardiological tests prescribed as part of their disease assessment. In reality, the course of the disease is highly variable and difficult to predict.

This is particularly true of their patients’ relatives who, despite having a similar genetic mutation, sometimes have a completely different clinical picture and/or evolution.

In fact, they are currently using a sudden death risk calculation proposed by the European Society of Cardiology (ESC) in 2014. The relevance of these risk scores is modest, and the distinction between high-risk and low-risk patients is insufficient and remains controversial.

It aims to establish, as accurately as possible and for each patient, a polygenic risk score (PRS), un score de risque polygénique (PRS), and integrate it with cardiological data to better identify patients who will benefit from pro-active therapeutic management, in particular via implantation of a medical device (endocavitary or subcutaneous automatic defibrillator, for example).

Artificial Intelligence (AI) will provide new risk stratification with direct translational application to guide therapeutic decision-making. In fact, we want it to enable clinicians to know the risk to which their patients are exposed, in order to adapt their decision-making and the frequency of their follow-up.

The CardI-HACK project will be managed by the Prof. Philippe Charron, Head of the Reference Center for Hereditary and Rare Heart Diseases, supported by Aurélie Foucher, Project Manager in the ICAN IHU Innovation and Research Valorization team.

Anonymizing genomic data is a key issue for the CardI-HACK project. Combining the analysis of specialists in artificial intelligence methods from Centre de la Sorbonne pour l’intelligence artificielle (SCAI) with the regulatory expertise of IHU ICAN and its data protection delegate, with the help of the regulatory and legal teams from Health Data HubA methodology for anonymizing genomic data will be defined to meet the regulatory requirements of the RGPD and the scientific requirements of the CardI-HACK project, and submitted to the CNIL for approval.

This project has received funding from Bpifrance as part of the “Data Challenges in Healthcare” call for projects under the France 2030 plan.

Financing obtained: €199,934