| Comparative Analysis of Sway Back and Anterior Pelvic Crossed Syndrome Using Support Vector Machine |
| Paper ID : 1343-SPORTCONGRESS |
| Authors |
|
آلا باباجان تبار1, فواد صیدی *2, سحر بیدختی نژاد3 1دانشجو ارشد،دانشکده علوم ورزشی و تندرستی،دانشگاه تهران،تهران،ایران 2استاد تمام،دانشکده علوم ورزشی و تندرستی،دانشگاه تهران،تهران 3دانشجو ارشد ،دانشکده مهندسی محیط زیست،دانشگاه تهران،تهران،ایران |
| Abstract |
| Abstract Introduction: Postural assessment is essential in corrective exercise and rehabilitation; however, differentiating between similar syndromes remains challenging. Sway-Back and Anterior Pelvic Crossed Syndrome (APCS) share multiple features, including pelvic displacement and reduced lumbar lordosis. These overlapping characteristics frequently lead to diagnostic ambiguity and classification uncertainty during clinical evaluation. In contrast, machine learning algorithms, such as the Support Vector Machine (SVM) (Noble, 2006), after appropriate training, enable precise and quantitative analysis. This study examines the similarities and differences between Sway-Back and APCS, demonstrating how SVM can identify subtle but decisive variations that define each syndrome. Methods: A conceptual dataset was constructed from scholarly articles and established textbooks (Key, 2010) (Sahrmann, 2001) (Saheb, Seidi, & Khezri, 2025) in the field of corrective exercise and used to train the model. Following training, the Support Vector Machine (SVM) identified both similarities and distinctions between these two postural syndromes, Sway-Back and Anterior Pelvic Crossed Syndrome (APCS). The model effectively distinguished postural patterns and quantified their subtle difference. Results: The SVM model revealed that many features are shared between Sway-Back and APCS. Testing on an individual clinically diagnosed with Sway-Back produced a classification result of 51.5% Sway-Back and 48.5% APCS. This indicates a high degree of feature overlap, affirming that subtle pelvic and trunk variations are key determinants in postural classification. These findings emphasize that minor variations, although small in percentage, can shift an individual’s posture from one syndrome to another. Conclusion: The Support Vector Machine provides a robust framework for distinguishing postural syndromes with overlapping features. Importantly, subtle differences in specific characteristics directly inform exercise prescription, as the corrective exercise approach for APCS differs from that for Sway-Back. Integrating AI-driven analysis into postural assessment enhances diagnostic accuracy, reduces inter-observer variability, and supports individualized, evidence-based interventions for postural correction and movement optimization. |
| Keywords |
| Postural Assessment, Sway-Back, Anterior Pelvic Crossed Syndrome, Support Vector Machine, Corrective Exercise, Machine Learning algorithms. |
| Status: Abstract Accepted (Poster Presentation) |
