Zainal Bakri, Siti Farhana (2023) A study on lung abnormity predictors among welders in the automotive manufacturing industries using a decision tree model. Doctoral thesis, Universiti Tun Hussein Onn Malaysia.

Preview	Text 24p SITI FARHANA ZAINAL BAKRI.pdf Download (1MB) | Preview
	Text (Copyright Declaration) SITI FARHANA ZAINAL BAKRI COPYRIGHT DECLARATION.pdf Restricted to Repository staff only Download (498kB) | Request a copy
	Text (Full Text) SITI FARHANA ZAINAL BAKRI WATERMARK.pdf Restricted to Registered users only Download (64MB) | Request a copy

Abstract

The risk prediction model has been applied and used widely in various fields, especially in clinical related studies. The transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD) has highlighted that it is crucial to find an important predictor before making a health risk prediction model. Welders are simultaneously exposed to multiple welding fume constituents at once, and the exposure to these fumes contributes to lung problems. To fill this gap, the important predictors of lung abnormalities due to welding fume exposure need to be investigated. This study explored the relationship between a demographic and heavy metal concentration found in the welding areas and the lung abnormalities of welders in automotive industries. A quantitative cross-sectional survey design is used. Individual samples of welding fumes were taken in Plants 1, 2, and 3 to determine the concentration of heavy metal constituents, along with a series of pulmonary function tests, biomarker samples from toenails, and a questionnaire on respiratory problems (exposed group = 48 welders, control group = 42 workers). Descriptive and correlation analyses were used to examine the data. A simple welding- lung assessment (WELA) tree model was created that explained the relationship among the independent variables. The model showed that the concentration of cobalt and aluminium were significant and made a unique contribution to the model as predictors of lung abnormalities in welders in validated test set (ROC/AUC: 0.66, 95% CI: 0.4167– 0.9033, sensitivity = 60%, specificity = 20%, precision = 80% and accuracy = 70%). These findings provide new insight into the assessment of lung risk caused by welding fumes using a simple classification tree approach. This groundbreaking study not only identified predictors of lung abnormalities among Malaysian automotive welders but also laid the foundation for a future risk-prediction model, revolutionizing the field of occupational health and safety

Actions (login required)

View Item