Integration of routine hematological indicators for early diagnosis of occupational silicosis

Objectives: To explore the value of a logistic regression model based on haematological parameters for the early diagnosis of silicosis by comparing the differences in haematological parameters between silicosis patients and healthy physical examiners. Methods: A total of 390 individuals, including 195 silicosis patients and 195 normal participants were included in the training cohort. Then, 65 silicosis patients and 65 healthy individuals were enrolled in the validation cohort. Whole blood samples were collected from all participants, and hematological indicator characteristics were determined. Features with statistical significance in the univariate analysis of the training cohort and reported significant features were included in the logistic regression analysis to determine the independent factors influencing the diagnosis of silicosis and to construct a logistic diagnostic model. A receiver operating characteristic (ROC) curve was plotted to evaluate the accuracy of the model in diagnosing silicosis. Results: In the training cohort, several hematological indicators were significantly different in silicosis patients, including Hematocrit(HCT), Hemoglobin(HGB), Mean corpuscular volume(MCV), Red Blood Cell Count(RBC), White blood cell count (WBC), Mon#, Mon%, Neu#, Neu%, Red blood cell distribution width coefficient of variation(RDW_CV), C-reactive protein(CRP), Hydroxybutyrate dehydrogenase (HBDH), Lactate dehydrogenase(LDH), Prothrombin time(PT), International normalized ratio(INR), Fibrinogen(FIB), and D-Dimer(DD) levels, all with statistical significance ( P  < 0.05). The silicosis diagnostic model performed well in the training cohort (Area Under Curve, AUC = 0.943) and had high diagnostic sensitivity (83.1%) and specificity (92.3%). The diagnostic model also effectively distinguished between silicosis patients and the control cohort in the validation cohort (AUC = 0.936). Conclusions: This study confirmed that Age, CRP, LDH, Macro%, and INR were independent factors influencing the diagnosis of silicosis, and the logistic regression model based on these indicators could provide a reliable basis for predicting silicosis diagnosis.