Redha Attaoua

This paper develops a multi-Polygenic Risk Score (multiPRS) to predict diabetes complications in different ethnic groups, finding that logistic regression with Mondrian Cross-Conformal Prediction improves prediction accuracy in non-European populations.

The paper introduces a hierarchical transformer model to detect genetic block interactions, showing improved performance over traditional methods in simulation and UK Biobank data.

This paper develops a polygenic risk score model to predict diabetic complications and their response to therapy using genome-wide association study data.