Causal and Federated Multimodal Learning for Cardiovascular Risk Prediction under Heterogeneous Populations

TL;DR

This paper introduces a comprehensive multimodal framework combining transformers, graph neural networks, and causal learning to predict cardiovascular risk with interpretability and privacy preservation across diverse populations.

Contribution

It presents a novel integrated model for CVD risk prediction that incorporates causal invariance, federated learning, and interpretability techniques, advancing clinical trustworthiness.

Findings

Achieves state-of-the-art discrimination and robustness across cohorts

Maintains fair performance across demographic groups

Ensures privacy-preserving and interpretable predictions

Abstract

Cardiovascular disease (CVD) continues to be the major cause of death globally, calling for predictive models that not only handle diverse and high-dimensional biomedical signals but also maintain interpretability and privacy. We create a single multimodal learning framework that integrates cross modal transformers with graph neural networks and causal representation learning to measure personalized CVD risk. The model combines genomic variation, cardiac MRI, ECG waveforms, wearable streams, and structured EHR data to predict risk while also implementing causal invariance constraints across different clinical subpopulations. To maintain transparency, we employ SHAP based feature attribution, counterfactual explanations and causal latent alignment for understandable risk factors. Besides, we position the design in a federated, privacy, preserving optimization protocol and establish rules for convergence, calibration and uncertainty quantification under distributional shift. Experimental studies based on large-scale biobank and multi institutional datasets reveal state discrimination and robustness, exhibiting fair performance across demographic strata and clinically distinct cohorts. This study paves the way for a principled approach to clinically trustworthy, interpretable and privacy respecting CVD prediction at the population level.