Research Information System
Mayo Clinic Proceedings: Digital Health, vol.4, no.1, 2026 (ESCI, Scopus)
Objective: To evaluate the effectiveness of the artificial intelligence–based qXR lung nodule malignancy score (qXR-LNMS) in detecting high-risk incidental pulmonary nodules (IPNs) on chest X-rays (CXRs). Patients and Methods: The CREATE (NCT05817110), a prospective, observational study for participants aged 35 years or older with IPN (size, ≥8 to ≤30 mm) on CXR, enrolled 712 participants (high-risk: 498 and low-risk: 214) between April 1, 2023, and December 31, 2024. Participants were flagged by the Food and Drug Administration–cleared qXR detection algorithm and confirmed by radiologists. Threshold for success was set at 20% for positive predictive value (PPV) and 70% for negative predictive value (NPV). The primary and secondary outcomes included PPV and NPV of qXR-LNMS against the risk of malignancy assessed by radiologists using low-dose computed tomography (LDCT) and binarized risk categories based on Lung-RADS score and Mayo Clinic model and PPVs and NPVs by clinicodemographic characteristics with 95% CIs using Wilson score method. Results: Overall, the PPV and the NPV of qXR-LNMS risk prediction against radiologists’ assessment on LDCT were 54.2% (95% CI, 49.8-58.5) and 93.5% (95% CI, 89.3-96.1), respectively. The agreement between Mayo Clinic model and qXR-LNMS was observed in 70.6% participants (Spearman correlation, 0.247). Results across key subgroups were consistent with all PPV and NPV point estimates crossing the prespecified threshold. Conclusion: The results demonstrate the potential of qXR-LNMS in predicting benign and malignant IPN on CXR, thereby supporting lung cancer screening, particularly in resource-limited settings, although further validation is needed. Trials Registration: clinicaltrials.gov Identifier: NCT05817110