Evaluation of the effect of dance therapy for children with high-functioning autism based on the LightGBM model

Abstract

In the context of big data development, the problem of evaluating the effectiveness of dance therapy for children with autism has attracted the attention of researchers. In this paper, feature selection and model construction use the Pearson correlation coefficient and sequence forward selection method to select the optimal feature subset through a Bayesian algorithm-based optimization of integrated LightGBM model, LightGBM model for Latin dance, aerobics, ballet, jazz, street dance, folk dance six autistic children treatment effect evaluation of weight and accuracy rate indicators. In the weight index, 93.4% of the improved LightGBM model performed better than 83.2% of the traditional LightGBM model. In the accuracy index, 95.6% of the improved LightGBM model performed better than 93.4% of the traditional LightGBM model. This study has a guiding reference value for treating autism in children, thus helping more autistic patients.