Machine learning models for minimizing aggravation in work-related musculoskeletal disorders among slaughterhouse workers.
Journal:
Work (Reading, Mass.)
Published Date:
Apr 28, 2025
Abstract
BackgroundWork-related musculoskeletal disorders (WMSDs) are common in Brazilian slaughterhouses. The repetitive and strenuous nature of meat processing, especially in slaughterhouses, makes employees highly susceptible to developing WMSDs. Prolonged standing, repetitive motions, and forceful actions such as lifting and cutting are common contributing factors.ObjectiveThis study aimed to develop models to predict the risk of aggravating WMSDs in slaughterhouse workers using the data mining concept.MethodsData were retrieved from an open-source governmental database, and descriptive statistics were used to evaluate them. The data set involved organizational aspects, and demographic, physical, and health issues were attributes. A descriptive analysis was applied, and the data mining method was used to process data with the Random Forest algorithm to classify the aggravation of WMSDs'.ResultsThree tree-ensemble predictive models were found (accuracy = 95.3%, κappa = 0.93) and described using the "If-Then" rules. The first tree had as the root attribute the change of function due to a health condition (high blood pressure or diabetes), followed by medical leave, working time, change of working place, and age, and the second had the worker's age as the root attribute, followed by working time, sex, and age. The third tree's root attribute was musculoskeletal pain symptoms, followed by working hours, age, and working time. Workers who do not change their roles and are on medical leave for over 1642.5 days present a high risk of worsening symptoms. Working time over 1980 days leads to a high risk of aggravating WMSDs. Females older than 24.5 years and staying more than 1620 days in the same function also presented a high risk of aggravating the WMSDs.ConclusionsThe machine learning models might help prevent WMSD risk aggravation by sorting the available data set and identifying patterns and relationships.
