The work aims to propose a prediction model of product emotional satisfaction based on comment data aiming at the complex mapping problem between multi-dimensional design elements and user emotional satisfaction of health care assistive devices. Firstly, the comment data of samples were crawled, the multi-dimensional element space of products was constructed by LDA model and the elements of the samples were coded. Secondly, the emotional satisfaction of the comment data corresponding to each sample was analyzed to obtain the emotional satisfaction score. Finally, the emotional satisfaction prediction model was constructed by BP neural network, and the reliability of the model was tested by K-fold cross-validation method. With the electric wheelchair as an example, the mean square error between the predicted value and the expected value was 0.044 1 after cross-validation of 30 sample data sets, which showed that the prediction model had good accuracy and high reliability. Next, the emotional satisfaction of 104 976 random combination schemes was calculated by the prediction model, and the best combination scheme of electric wheelchair was obtained. The prediction model of product emotional satisfaction based on user comment data can effectively establish the matching relationship between multi-dimensional design elements and emotional satisfaction, assist designers to quickly identify the combination of multi-dimensional design elements with high user emotional satisfaction, and improve the scientific decision-making of design schemes.