Objective:To explore the quantitative indexes of dual energy spectrum CT and related clinical data to establish a predictive model for predicting pneumoconiosis.Methods:In April 2024, the information of 203 pneumoconiosis patients diagnosed by the occupational disease appraisal expert group in the Third People's Hospital of Xinjiang Uygur Autonomous Region (Occupational Disease Hospital of Xinjiang Autonomous Region) from January 2022 to December 2023 was retrospectively analyzed. Another 207 non-pneumoconiosis patients with dust exposure history were selected as control group. The measurement data between the two groups were compared using T test or Wilcoxon in dependent quality test, count date asing chi-square or Fishers test, the energy spectrum related indicators and clinical indicators of the patients were compared between groups, and potential factors for diagnosis of pneumoconiosis were screened through univariate analysis, and independent risk factors were further determined by multivariate logistic regression. Based on the results of regression analysis, the machine learning model was constructed, and the reciver operating characteristic curve (ROC) was drawn to evaluate the efficiency of the model, and the Area under cruve (AUC) value, sensitivity and specificity were calculated.Results:Smoking, lung tissue mass, silicon dioxide (SiO 2) equivalent total mass and SiO 2 equivalent concentration were the risk factors for pneumoconiosis ( P<0.05) . Multivariate logistic regression analysis showed that smoking, lung tissue mass, total lung SiO 2 equivalent total volume and total lung SiO 2 equivalent total mass were independent predicators of the diagnosis of pneumoconiosis ( OR=0.53, 0.99, 1.13, 0.85, P<0.05) . Logistic regression machine learning was used to establish a predictive model, and the training set AUC was 0.74, and the verification set AUC was 0.72, indicating that the model had good accuracy and certain ability to diagnose pneumoconiosis. Conclusion:The machine learning prediction model established by the quantitative analysis index of dual energy spectrum CT and clinical related indexes has a good diagnostic performance for the diagnosis of pneumoconiosis.

Objective:To explore the quantitative indexes of dual energy spectrum CT and related clinical data to establish a predictive model for predicting pneumoconiosis.Methods:In April 2024, the information of 203 pneumoconiosis patients diagnosed by the occupational disease appraisal expert group in the Third People's Hospital of Xinjiang Uygur Autonomous Region (Occupational Disease Hospital of Xinjiang Autonomous Region) from January 2022 to December 2023 was retrospectively analyzed. Another 207 non-pneumoconiosis patients with dust exposure history were selected as control group. The measurement data between the two groups were compared using T test or Wilcoxon in dependent quality test, count date asing chi-square or Fishers test, the energy spectrum related indicators and clinical indicators of the patients were compared between groups, and potential factors for diagnosis of pneumoconiosis were screened through univariate analysis, and independent risk factors were further determined by multivariate logistic regression. Based on the results of regression analysis, the machine learning model was constructed, and the reciver operating characteristic curve (ROC) was drawn to evaluate the efficiency of the model, and the Area under cruve (AUC) value, sensitivity and specificity were calculated.Results:Smoking, lung tissue mass, silicon dioxide (SiO 2) equivalent total mass and SiO 2 equivalent concentration were the risk factors for pneumoconiosis ( P<0.05) . Multivariate logistic regression analysis showed that smoking, lung tissue mass, total lung SiO 2 equivalent total volume and total lung SiO 2 equivalent total mass were independent predicators of the diagnosis of pneumoconiosis ( OR=0.53, 0.99, 1.13, 0.85, P<0.05) . Logistic regression machine learning was used to establish a predictive model, and the training set AUC was 0.74, and the verification set AUC was 0.72, indicating that the model had good accuracy and certain ability to diagnose pneumoconiosis. Conclusion:The machine learning prediction model established by the quantitative analysis index of dual energy spectrum CT and clinical related indexes has a good diagnostic performance for the diagnosis of pneumoconiosis.