An investigation of musculoskeletal discomforts among mining truck drivers with respect to human vibration and awkward body posture using random forest algorithm

Using Random Forest algorithms, this study aimed to investigate musculoskeletal discomforts among mining truck drivers considering human vibration and awkward body posture. The study was conducted on 65 professional male drivers of mining trucks. The Cornell questionnaire was used to determine musculoskeletal discomforts. Drivers' exposure to vibrations was measured using the Svanteck 106 A vibration meter. The body posture was analyzed using the quick exposure check (QEC). The main mechanical and individual risk factors were used as predictor variables of musculoskeletal discomforts model. The relative importance of each feature on the discomforts was determined based on Random Forest algorithm compared with multiple linear regression using R Statistics Packages. The equivalent acceleration of whole-body vibration (WBV) was higher than the exposure limit, however, the equivalent acceleration of hand-transmitted vibration (HTV) was lower than the exposure limit. The body posture of drivers was from moderate to high risk so that investigation and changes are required soon. The predictive error of Random Forest model for musculoskeletal discomfort scores was at an acceptable level with root mean square error (RMSE) = 5.29 for the blind case of drivers compared with regressions model with RMSE = 15.92. Random forest showed that the awkward body posture, vibration, and age, respectively, have the greatest relative importance on musculoskeletal discomforts. The findings provide empirical evidence on the relative importance of risk factors on musculoskeletal discomfort so that awkward body posture has a greater effect compared with whole-body vibration. Random forest provided better outputs and was more accurate compared with the regression method.