Machine learning analysis of cervical balance in early-onset scoliosis post-growing rod surgery: a case-control study

Abstract

Abstract We aimed to analyze the cervical sagittal alignment change following the growing rod treatment in early-onset scoliosis (EOS) and identify the risk factors of sagittal cervical imbalance after growing-rod surgery of machine learning. EOS patients from our centre between 2007 and 2019 were retrospectively reviewed. Radiographic parameters include the cervical lordosis (CL), T1 slope, C2-C7 sagittal vertical axis (C2-7 SVA), primary curve Cobb angle, thoracic kyphosis (TK), C7-S1 sagittal vertical axis (C7-S1 SVA) and proximal junctional angle (PJA) were evaluated preoperatively, postoperatively and at the final follow-up. The parameters were analyzed using a t-test and χ2 test. The machine learning methodology of a sparse additive machine (SAM) was applied to identify the risk factors that caused the cervical imbalance. 138 patients were enrolled in this study (96 male and 42 female). The mean thoracic curve Cobb angle was 67.00 ± 22.74°. The mean age at the first operation was 8.5 ± 2.6yrs. The mean follow-up was 38.48 ± 10.87 months. CL, T1 slope, and C2-7 SVA increased significantly in the final follow-up compared with the pre-operative data. (P 0.05). According to the SAM analysis of machine learning algorithms, Postoperative PJK, more improvement of kyphosis, and T1 slope angle were identified as the risk factors of cervical sagittal imbalance during the treatment of growing rod surgery. The growing rod surgery in EOS significantly affected the cervical sagittal alignment. Postoperative PJK and more improvement of kyphosis and T1 slope angle would lead to a higher incidence of cervical sagittal imbalance.

Keywords

• Cervical alignment
• Early-Onset Scoliosis
• Growing rod
• Proximal junctional kyphosis
• Machine learning algorithm