Genomic Prediction Models for Celiac Disease Risk Across Six European Cohorts

Gad Abraham developed a genomic risk score (GRS) model for celiac disease using statistical learning on genome-wide SNP profiles from six European cohorts. The model achieved cross-validation AUCs of 0.87–0.89 and independent replication AUCs of 0.86–0.9, explaining 30–35% of disease variance. This proof-of-concept demonstrates the potential for genomic-based risk stratification to improve clinical diagnostic pathways.

Use Cases

• Developing genomic risk scores for celiac disease based on genome-wide SNP profiles.
• Stratifying patients into higher-risk categories for definitive testing based on the described GRS.
• Comparing the performance of statistical learning approaches to HLA typing for disease exclusion.
• Adapting predictive model thresholds for different clinical scenarios as mentioned in the description.

Strengths

• Model performance was validated across six independent European cohorts.
• Achieved high predictive accuracy with AUCs between 0.86 and 0.9 in independent replication.
• The genomic risk score demonstrated a negative predictive value of ≥99.6% for excluding celiac disease.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic bias inherent to the six European cohorts used.

Provenance

Source

Gad Abraham

Collection Method

Statistical learning applied to genome-wide SNP profiles from six European celiac disease cohorts.

Geography

European cohorts (specific countries not detailed)