Analysis of Sleep Apnea Detection Based on U-Net and SAM Module
This paper describes a U-Net model based on pulse oxygen saturation(SpO2),which extracts multi-level features of blood oxygen signals through an encoding decoding structure.It uses the SAM module to adaptively increase the weight of oxygen reduction events.Cross validation was performed on the SleepApnea ECG dataset,with an average accuracy,sensitivity,and specificity of 95.69％,95.52％,and 95.59％,respectively,indicating higher detection accuracy compared to other models.
NETL
NSTL
万方数据
