Hypothesis: Pooling QuantiFERON-TB Gold IGRA quantitative values across multiple rheumatology centers using fully homomorphic encryption (FHE) will enable derivation of immunosuppression-adjusted diagnostic thresholds that improve LTBI detection sensitivity by >=15% in patients on biologic therapy, without exposing individual patient data.

Background: Standard IGRA uses fixed cutoff (>=0.35 IU/mL), but immunosuppressed patients on rituximab, JAK inhibitors, or high-dose glucocorticoids have impaired IFN-gamma production, yielding false-negative rates of 10-30% (Bartalesi et al., J Rheumatol 2018). Multi-site pooling could derive adjusted cutoffs but is blocked by HIPAA/GDPR constraints.

Approach: Using TFHE or CKKS FHE schemes, encrypted IGRA quantitative values, immunosuppression profiles (drug, dose, duration), and TB outcomes are aggregated across >=5 centers. Logistic regression on encrypted data derives optimal cutoffs stratified by immunosuppression level — no site decrypts another sites data.

Testable prediction: FHE-pooled adjusted IGRA thresholds (e.g., >=0.20 IU/mL for rituximab-treated patients) will detect >=15% more true LTBI cases vs standard 0.35 IU/mL cutoff, validated by TB reactivation events over 24-month follow-up.

Clinical implication: First privacy-preserving, multi-site LTBI screening optimization for immunosuppressed populations — systematically underserved by fixed-threshold diagnostics.

References: Bartalesi J Rheumatol 2018;45:1290. Lewinsohn Clin Infect Dis 2017;64:e1. Chillotti J Cryptol 2020;33:1235 (TFHE). Winthrop Ann Rheum Dis 2016;75:1133.