ATBdiscrimination

ATBdiscrimination classifies active tuberculosis (ATB) versus latent tuberculosis infection (LTBI) using 36 blood biochemical parameters and T-SPOT.TB detection results.

Key Features:

• Data Integration: Integrates 36 blood biochemical parameters with T-SPOT.TB detection results from routine clinical practice.
• Algorithmic Approach: Uses the random forest algorithm to develop a classifier distinguishing ATB from LTBI.
• Performance Metrics: Demonstrates AUC = 0.9256 on cross-validation and AUC = 0.8731 on external validation.

Scientific Applications:

• Clinical Diagnosis: Identification of active TB from routine blood tests and T-SPOT.TB results to inform patient management.
• Research Insights: Investigation of associations between tuberculosis infection status and routine blood biochemical parameters plus T-SPOT.TB results.

Methodology:

Model trained on data from 478 patients comprising 36 blood biochemical parameters and T-SPOT.TB results using a random forest classifier evaluated by cross-validation and external validation (AUCs 0.9256 and 0.8731, respectively).