[Machine learning analysis of the influence of workload factors and social psychological factors on WMSDs].

To explore the effects of workload factors and social psychological factors on work-related musculoskeletal disorders (WMSDs) , construct a preventive decision-assisted ensemble learning model, and propose screening methods with clinical significance. In October 2022, 1071 workers from optoelectronic enterprises were selected as the research subjects by cluster sampling method. The general situation of workers, workload factors, social psychological factors and the occurrence of WMSDs were collected by using questionnaires. logistic regression, Extreme Gradient Boosting (XGBoost) , ensemble learning and classification chain model were adopted to explore the key factors influencing WMSDs, and the area under curve (AUC) was used to evaluate the model performance. The incidence of WMSDs among workers in optoelectronic enterprises in the past year was 47.7% (511/1071) , among which the incidence of multi-site WMSDs was 54.4% (278/511) . logistic regression analysis showed that prolonged sitting, personnel shortage and forward neck tilt were risk factors for the occurrence of WMSDs in workers (<0.05) . XGBoost identified the key social psychological factors influencing WMSDs as low mood, mental tension, perceived happiness level, psychological calmness and tranquility. The integrated classification chain model based on the ordered label order had certain efficacy (AUC>0.7) when analyzing WMSDs at the neck, waist, shoulder, back, elbow and hip positions. Workload factors are the main risk factors for the occurrence of WMSDs among workers in optoelectronic enterprises, and social psychological factors also have potential influences. The construction of a classification chain model can accurately identify the occurrence of WMSDs in multiple parts. The alternating prevention strategy of workload factors and social psychological factors has important public health significance.