WP2 objectives: 1) Create anonymised repositories of raw data from CMR, echocardiography, and ECG collected during routine healthcare, paired with genomic data and clinical and demographic data collected in WP1; 2) Construct AI survival models.
Routine clinically-acquired imaging data (cardiac MRI, transthoracic echocardiography and digital ECGs) will be retrieved and pseudo-anonymised locally at each centre and imaging files stored in a central server. We have demonstrated that deep learning frameworks can improve survival predictions using similar imaging datasets in other cardiovascular disease and will apply these approaches to DCM. WP2 will also focus on optimizing feature selection from imaging and ECG data for survival models. These will then be combined with genomics for full multimodal analysis in WP5.