predicting work-related musculoskeletal disorders in indian construction workers using machine learning and deep learning classifiers

Background: construction workers often experience work-related musculoskeletal disorders (msds) at a high rate. the poor performance of workers due to its presence is a serious concern to all the stakeholders and it is necessary to diagnose before it develops. the study aimed to ascertain the performance of machine learning (ml) classifiers and multi-layer perceptron (mlp) neural networks in predicting msds.materials and methods: the cross-sectional study utilized the data on potential msd risk factors collected from 1040 construction workers on infrastructure projects across different states in india. the data was gathered through direct interactions with the construction workers and also, through the health records maintained by the safety department of the project sites. stratified random sampling was the approach used for sampling. the prediction of the development of msds is based on nine features. in this study, naive bayes (nb), k-nearest neighbors (knn), and xgboost classifiers were applied to predict the presence of msds, and the results were compared with the mlp neural network based on the metrics. results: in predicting the presence of msds, xgboost’s classifier, with 91% accuracy, was superior to nb, knn, and mlp neural networks having 87%, 72%, and 85% accuracy, respectively. a powerful prediction tool has been developed to diagnose msds and effectively interpret the outcomes confidently.conclusions: the performance metrics of the xgboost classifier resulted in the best compared with the other classifiers. the prediction tool is useful to diagnose the prevalence of msds in the early stages.