Blood Marker-Based Machine Learning Model for Survival Prediction in Patients with Pneumoconiosis: Construction and...

This study aimed to investigate the association between blood inflammatory markers, including C-reactive protein (CRP) and interleukin-6 (IL-6), and the short-term prognosis of pneumoconiosis, and to develop a multifactorial prediction model.

Clinical data of 813 pneumoconiosis patients admitted to two regional tertiary hospitals from October 2016 to August 2023 were retrospectively collected and randomly divided into a training set (n=568) and an external validation set (n=245). Variables were screened by least absolute shrinkage and selection operator (LASSO) regression, a multifactorial logistic regression column-line graph model was constructed, and the relationship between risk factors and prognosis was analyzed by multifactorial Cox regression and Kaplan-Meier survival curves. The model performance was verified by consistency index (C index), receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA) and clinical impact curve (CIC).

The mean age was 62.1 years in the training set and 63.3 years in the validation set, with mortality rates of 11.1% and 11.9%, respectively. LASSO regression identified age, dust exposure duration, dyspnea, blood oxygen saturation (SpO₂), neutrophil count, CRP, and IL-6 as predictors. CRP (OR=1.028, 95% CI: 1.016-1.040) and IL-6 (OR=1.020, 95% CI: 1.010-1.030) were independent risk factors for poor short-term prognosis (both P<0.001), consistent with Cox analysis. The model demonstrated excellent discrimination with AUCs of 0.905 in the training set and 0.920 in the validation set. Calibration showed good agreement between predicted and observed risks (Hosmer-Lemeshow P>0.05). DCA and CIC indicated high clinical value. High CRP/IL-6 levels were associated with significantly reduced survival (log-rank P<0.001).

The prediction model based on CRP, IL-6, and clinical characteristics effectively identifies pneumoconiosis patients at high risk of short-term poor prognosis, providing a reliable basis for early intervention due to its high discriminatory power and clinical applicability.