Multimodal Deep Learning for Pulmonary Nodule Detection on Chest Radiography in High-Risk Adults, With Secondary Validation for All-Cause and Cause-Specific Mortality Prediction: A Multicenter Cohort Study

Abstract

Chest radiographs (CXRs) may encode prognostic signals beyond pulmonary nodule detection. We developed LungProNet, a multimodal deep-learning (DL) model that fuses CXR features with four epidemiologic variables (age, sex, smoking history, and family history) for pulmonary nodule detection as the primary task, with secondary validation for all-cause and cause-specific mortality prediction. LungProNet was trained and internally validated on Tianjin Lung Cancer Imaging Dataset (TLCID) (70/30; n = 2852/1227) and externally validated on ChestDR (n = 4848), with stratified analyses across epidemiologic strata. Discrimination was quantified by area under the curve (AUC) (95% confidence intervals), with accuracy, sensitivity, and specificity reported, and results were benchmarked against contemporary machine learning/DL baselines. The pretrained multimodal encoder was transferred without fine-tuning to the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO) (n = 24,697); its fused embeddings were used as covariates in Cox proportional-hazards models, and time-dependent AUCs were evaluated at 1–12 years. For nodule detection, AUCs were 0.979 (0.975–0.982) in TLCID and 0.849 (0.835–0.862) in ChestDR; the TLCID stratified model reached 0.990 (0.984–0.994). In PLCO, AUCs were 0.925 (0.892–0.952) for all-cause mortality and 0.939–0.985 for cardiac-, lung cancer-, and Chronic Obstructive Pulmonary Disease (COPD)-cause mortality, with robust subgroup performance. These results support CXR-based nodule flagging within screening workflows and suggest secondary opportunistic risk stratification potential.

Keywords

chest X-ray / deep learning / lung cancer screening / mortality prediction / pulmonary nodule

References

[1] H. Sung, J. Ferlay, R. L. Siegel, et al., “Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA: A Cancer Journal for Clinicians 71, no. 3 (2021): 209–249.

[2] H. Zeng, W. Chen, and R. Zheng, “Changing Cancer Survival in China During 2003-15: A Pooled Analysis of 17 Population-Based Cancer Registries,” Lancet Global Health 6, no. 5 (2018): e555–e567.

[3] H. Zeng, X. Ran, L. An, et al., “Disparities in Stage at Diagnosis for Five Common Cancers in China: A Multicentre, Hospital-Based, Observational Study,” Lancet Public Health 6, no. 12 (2021): e877–e887.

[4] Y. Sim, M. J. Chung, E. Kotter, et al., “Deep Convolutional Neural Network-Based Software Improves Radiologist Detection of Malignant Lung Nodules on Chest Radiographs,” Radiology 294, no. 1 (2020): 199–209.

[5] P. Rajpurkar, J. Irvin, R. L. Ball, et al., “Deep Learning for Chest Radiograph Diagnosis: A Retrospective Comparison of the CheXNeXt Algorithm to Practicing Radiologists,” Plos Medicine 15, no. 11 (2018): e1002686.

[6] D. Ardila, A. P. Kiraly, S. Bharadwaj, et al., “End-to-End Lung Cancer Screening With Three-Dimensional Deep Learning on Low-Dose Chest Computed Tomography,” Nature Medicine 25, no. 6 (2019): 954–961.

[7] D. A. Rosman, R. Duszak, W. Wang, et al., “Changing Utilization of Noninvasive Diagnostic Imaging Over 2 Decades: An Examination Family-Focused Analysis of Medicare Claims Using the Neiman Imaging Types of Service Categorization System,” Ajr American Journal of Roentgenology 210, no. 2 (2018): 364–368.

[8] J. Chamberlin, M. R. Kocher, J. Waltz, et al., “Automated Detection of Lung Nodules and Coronary Artery Calcium Using Artificial Intelligence on Low-Dose CT Scans for Lung Cancer Screening: Accuracy and Prognostic Value,” BMC Medicine 19, no. 1 (2021): 55.

[9] L. He, H. Yu, L. Shi, et al., “Equity Assessment of the Distribution of CT and MRI Scanners in China: A Panel Data Analysis,” International Journal for Equity in Health 17, no. 1 (2018): 157.

[10] M. F. Bell, T. P. Jernigan, and R. S. Schaaf, “Prognostic Significance of Calcification of the Aortic Knob Visualized Radiographically,” American Journal of Cardiology 13 (1964): 640–644.

[11] J. N. Cohn, G. R. Johnson, R. Shabetai, et al., “Ejection Fraction, Peak Exercise Oxygen Consumption, Cardiothoracic Ratio, Ventricular Arrhythmias, and Plasma Norepinephrine as Determinants of Prognosis in Heart Failure. The V-HeFT VA Cooperative Studies Group,” Circulation 87, no. supplement, 6 (1993): Vi5–16.

[12] J. G. Nam, S. Park, E. J. Hwang, et al., “Development and Validation of Deep Learning-Based Automatic Detection Algorithm for Malignant Pulmonary Nodules on Chest Radiographs,” Radiology 290, no. 1 (2019): 218–228.

[13] J. T. Wu, H. A. Wakelee, and S. S. Han, “Optimizing Lung Cancer Screening with Risk Prediction: Current Challenges and the Emerging Role of Biomarkers,” Journal of Clinical Oncology 41, no. 27 (2023): 4341–4347.

[14] L. Guo, “Lung Cancer Risk Prediction Nomogram in Chinese Female Non-smokers,” Journal of Thoracic Oncology 17, no. supplement, 9 (2022): S35–S36.

[15] M. Oudkerk, A. Devaraj, R. Vliegenthart, et al., “European Position Statement on Lung Cancer Screening,” Lancet Oncology 18, no. 12 (2017): e754–e766.

[16] A. McWilliams, M. C. Tammemagi, J. R. Mayo, et al., “Probability of Cancer in Pulmonary Nodules Detected on First Screening CT,” New England Journal of Medicine 369, no. 10 (2013): 910–919.

[17] C. I. Henschke, D. F. Yankelevitz, D. M. Libby, et al., “Survival of Patients With Stage I Lung Cancer Detected on CT Screening,” New England Journal of Medicine 355, no. 17 (2006): 1763–1771.

[18] C. de Margerie-Mellon and G. Chassagnon, “Artificial Intelligence: A Critical Review of Applications for Lung Nodule and Lung Cancer,” Diagnostic and Interventional Imaging 104, no. 1 (2023): 11–17.

[19] Z. Gandhi, P. Gurram, B. Amgai, et al., “Artificial Intelligence and Lung Cancer: Impact on Improving Patient Outcomes,” Cancers 15, no. 21 (2023): 5236.

[20] J. G. Nam, S. Park, E. J. Hwang, et al., “Development and Validation of Deep Learning–Based Automatic Detection Algorithm for Malignant Pulmonary Nodules on Chest Radiographs,” Radiology 290, no. 1 (2019): 218–228.

[21] S. Schalekamp, B. van Ginneken, E. Koedam, et al., “Computer-aided Detection Improves Detection of Pulmonary Nodules in Chest Radiographs Beyond the Support by Bone-Suppressed Images,” Radiology 272, no. 1 (2014): 252–261.

[22] Y. Sim, M. J. Chung, E. Kotter, et al., “Deep Convolutional Neural Network–Based Software Improves Radiologist Detection of Malignant Lung Nodules on Chest Radiographs,” Radiology 294, no. 1 (2020): 199–209.

[23] M. T. Lu, A. Ivanov, T. Mayrhofer, et al., “Deep Learning to Assess Long-Term Mortality From Chest Radiographs,” JAMA Network Open 2, no. 7 (2019): e197416.

[24] J. Weiss, V. K. Raghu, D. Bontempi, et al., “Deep Learning to Estimate Lung Disease Mortality From Chest Radiographs,” Nature Communications 14, no. 1 (2023): 2797.

[25] J. Weiss, V. K. Raghu, K. Paruchuri, et al., “Deep Learning to Estimate Cardiovascular Risk From Chest Radiographs: A Risk Prediction Study,” Annals of Internal Medicine 177, no. 4 (2024): 409–417.

[26] M. M. Oken, W. G. Hocking, P. A. Kvale, et al., “Screening by Chest Radiograph and Lung Cancer Mortality: The Prostate, Lung, Colorectal, and Ovarian (PLCO) Randomized Trial,” JAMA 306, no. 17 (2011): 1865–1873.

[27] P. F. Pinsky, A. Miller, B. S. Kramer, et al., “Evidence of a Healthy Volunteer Effect in the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial,” American Journal of Epidemiology 165, no. 8 (2007): 874–881.

RIGHTS & PERMISSIONS

2026 The Author(s). MedComm published by Sichuan International Medical Exchange & Promotion Association (SCIMEA) and John Wiley & Sons Australia, Ltd.