A novel serum four-biomarker biosignature and field-deployable colloidal gold strip for accurate identification of subclinical and active tuberculosis.

BACKGROUND: Current TB diagnostics fail to detect subclinical and paucibacillary cases, hindering End TB Strategy goals. We validated a novel serum four-biomarker biosignature and developed a colloidal gold strip for accurate subclinical and active TB identification. METHODS: A total of 280 participants (HC, PN, STB, ATB; n = 70 each) were enrolled. Serum PDGFC, CXCL9, CXCL11, and CCL1 were measured by ELISA in training (n = 200) and external validation (n = 80) cohorts. Six machine‑learning algorithms built diagnostic models. A four-plex colloidal gold strip with simplified visual criteria was developed and compared against smear, TB DNA, and IGRA. RESULTS: All biomarkers were significantly elevated in STB/ATB versus HC/PN (p < 0.001), confirming Mtb-specific responses. Logistic regression achieved 97.5% validation sensitivity. The strip using a '≥2 positive lines' criterion yielded 96.0% sensitivity, 95.0% specificity, 95.5% accuracy, and outperformed conventional tests, effectively identifying smear-negative subclinical TB. CONCLUSIONS: This biosignature is a robust TB diagnostic tool. The companion strip offers a rapid, sputum-free, instrument-minimal solution for decentralized screening, bridging biomarker discovery and field application in resource-limited settings.