<p>Traditional sports injury monitoring methods include video analysis, fixed force platforms, and optical motion capture systems, which have limited accuracy, poor real-time performance, and insufficient portability, making efficient and accurate high-risk movement classification difficult. This paper introduced a sports injury monitoring solution based on high-performance wearable sensors. By integrating flexible strain sensors, it achieved a surface electromyography sensors and inertial measurement units, synchronous and highly sensitive acquisition of multimodal physiological signals. This paper also used a low-power Bluetooth to achieve wireless high-speed data transmission, and introduced fast feature extraction (sliding window root mean square (RMS), sample entropy (SampEn)) and lightweight convolutional neural network models to solve the problem of poor real-time performance. It used bandpass filtering and sliding window feature extraction methods to improve signal processing efficiency, and combined lightweight convolutional neural networks to achieve accurate real-time identification of motion status and high-risk movement patterns. The experimental results show that the signal-to-noise ratio of the collected signal is maintained above 20 dB, and the transmission data loss rate does not exceed 10%. The solution has shown excellent accuracy, response speed and wearing comfort in various sports scenarios, providing effective technical support for sports training and high-risk movement assessment.</p>