Dottorato in TECNOLOGIE BIOMEDICHE INNOVATIVE IN MEDICINA CLINICA

Titolo della tesi: Evaluating Cardiac Phenotype in 22q11.2 Deletion Syndrome Patients: AI-Enhanced Characterization

22q11.2 deletion syndrome (22q11.2DS) exhibits a heterogeneous clinical presentation and complex genotype-phenotype correlations. While 80% of individuals display congenital heart defects, even non-cardiac-deleted patients show an increased risk of sudden death from unknown causes. This underscores the need of in-depth phenotypic analysis in the management of these patients. Our study aims to characterize the cardiac phenotype in individuals with 22q11.2DS using a multidisciplinary approach centered on the establishment of a database incorporating clinical, genetic, and imaging data. Subsequent analysis involves advanced techniques, including artificial intelligence (AI). Of the total 110 patients monitored at our center, we examined 34 patients with 22q11.2DS, defining phenotypic classes using the Human Phenotype Ontology (HPO) system. For each patient with 22q11.2DS and a control group of 137 patients, we analyzed electrocardiogram (ECG) data. In collaboration with the Department of Automatic and Management Engineering, we developed a Python tool for the automatic analysis of ECG traces and created a structured dataset linking ECG data to HPO classes of patients. Statistical analysis using the Mann-Whitney test revealed a significant reduction in the ST segment (indicative of ventricular repolarization) in 22q11.2DS patients without congenital heart defects compared to non-deleted controls without CHD. These findings were further validated through a machine learning model based on a decision tree. Despite the limited number of patients, the created dataset could be utilized to train artificial intelligence models to automate ECG annotation and identify early differences between patients with and without 22q11.2DS, using a widely adopted instrumental test such as the ECG.