aceHarnessing Exhaled Breath for Lung Cancer Early Detection-Results From the ExPeL Study
Scalable, non-invasive tools are critically needed to improve early lung cancer detection and optimize primary care referral pathways. We evaluated Inflammacheck, a point-of-care device utilizing exhaled breath condensate (EBC) H 2 O 2 and physiological parameters with machine learning for non-invasive lung cancer detection in a real-world screening population. Exhaled Hydrogen Peroxide for Early Lung Cancer Detection (ExPeL) study participants, from the UK Targeted Lung Health Check (TLHC) programme, included i…
Pulmonary adenocarcinoma, left lung, CT scan Case 239 (7603361542) by Yale Rosen from USA. CC BY-SA 2.0 · https://creativecommons.org/licenses/by-sa/2.0
Researchers tested Inflammacheck, a point-of-care device that measures hydrogen peroxide in exhaled breath condensate plus physiological signals, combined with machine learning, in 34 participants from a UK lung health check programme where 83% of cancers were stage I-II. Multivariate analyses separated cancer and control groups, and a voting ensemble achieved 85.7% accuracy and 0.90 ROC-AUC on held-out data.
The result matters because scalable non-invasive triage tools are needed for primary care where symptoms are non-specific and prevalence is low. Whether the 100% specificity and high accuracy persist beyond this small, early-stage-enriched cohort, and how metabolomic signatures integrate with the point-of-care sensor, remains to be validated in larger prospective screening workflows.
- ExPeL study enrolled participants from the UK Targeted Lung Health Check programme including suspected lung cancer cases and low-risk ever-smoker controls.
- Device measured H2O2, end-tidal CO2, humidity, temperature, and exhalation flow rate with multivariate PCA, LDA and Mahalanobis analysis showing group separation.
- Untargeted LC-MS metabolomics identified 2132 features with four key metabolites yielding AUC 0.969 for cancer discrimination.
A point-of-care exhaled breath condensate device combined with physiological parameters and ensemble machine learning identified early-stage lung cancer in a real-world screening cohort with 85.7% accuracy and 100% specificity on held-out test data.
The rundown
The study collected exhaled breath condensate via Inflammacheck in the ExPeL cohort drawn from the UK Targeted Lung Health Check programme. In addition to H2O2, the device captured end-tidal CO2, humidity, temperature and flow rate, and analyses noted greater dispersion in cancer patients reflecting physiological heterogeneity missed by univariate analysis.
Supervised models were trained on SMOTE-balanced data and evaluated on held-out test sets, with the voting ensemble reporting no false positives. Parallel untargeted metabolomics was performed to identify discriminatory molecular features beyond the sensor parameters.
Sources
- Peer-reviewedClinical and Translational Science2026-08-01
How should this claim be treated?
The debate