Evaluation of factors associated with adult skeletal fluorosis in coal-burning type of endemic fluorosis and initial screening model based on machine learning in Guizhou, Southwest China

Abstract

Skeletal fluorosis caused by coal-burning type endemic fluorosis greatly affects the health of the population in the affected areas, but large-scale diagnostic work is limited by human and material resources, making it difficult to implement comprehensively. In this study, we investigate adults in coal-burning type endemic skeletal fluorosis areas in Guizhou. The study areas are selected by a comprehensive analysis of the detection rate of dental fluorosis in children aged 8–12 years in Guizhou for the year 2023. We collect information from questionnaires, physical examinations, and diagnostic X-ray Findings of Skeletal Fluorosis (XRF) in adults. The effective number of people investigated in this study was 2276, and the detection rate of XRF was 79.35 %. The top 5 relevant factors for skeletal fluorosis were age, educational background, height, Mini-Mental State Examination (MMSE) score and family population. Among the 8 models, random forest performed the best with an accuracy of 86.00 %, and the performance was more stable in the prevalence of different sizes, which provides a new idea for the prevention and treatment of skeletal fluorosis in coal-burning type of endemic fluorosis. In this study, the screening of the main correlates of XRF can provide an effective reference for the initial screening of skeletal fluorosis, and the practical application value of machine learning in this research field can be further explored.