Spinal health problems have attracted much attention, and the research of auxiliary diagnosis for patients with scoliosis has become a hot topic in the medical field. In this study, the combination of surface electromyography (sEMG) and machine learning (ML) was used to identify the spinal posture, so as to provide reference for the future diagnosis and screening of patients with scoliosis. The study included 15 participants and 540 data sets, and the accuracy of normal, right and left spine posture was more than 89%. This study has important reference value in the auxiliary diagnosis of scoliosis.