School Scoliosis Screening with Just a Smartphone — Safe, Simple, Smart, and Scalable!
This project revolutionizes adolescent scoliosis screening through TokenSTFormer—a cutting-edge AI model that analyzes gait videos captured by standard smartphones. Unlike error-prone manual exams that require participants to undress and maintain a static forward-bending posture, our system extracts 238 biomechanical featuresto construct a privacy-preserving Kinematic Knowledge Map. The model’s breakthrough Spatial-Temporal Tokenization (STT) separates motion dynamics into specialized semantic tokens, enabling unprecedented accuracy (79%) and sensitivity (84.5%) validated against clinical X-rays.
Exhibiting the world’s first dataset pairing 1,516 gait videos with golden-standard X-rays, ScoliDetect™ demonstrates how accessible technology enables cost-effective, high-quality, and scalable school/community screening. Its mobile compatibility and non-invasive design position it as the future of early spinal deformity detection.
Protected by Intellectual Property Rights
Unauthorized use or reproduction prohibited.
Patents & Copyrights:
- 一种疾病预测模型的构建及应用方法、模型、设备、介质
CN 119626507 A, CN 119626510 A, CN 119480082 A, CN 119626509 A, CN 119480081 A
- 一种医疗数据管理设备、电子设备、机器人
CN 119622806 A
- 疾病预测模型的构建方法及应用系统、模型、设备、介质
CN 119673423 A, PCT/CN2025/107777
- ScoliDetect系统
软著登字第15015857号
