The invention discloses a sleep apnea syndrome detection system based on LSTM neural network classification. The sleep apnea syndrome detection system is composed of a protection circuit, an impedancerespiration detection module, an electrocardiosignal detection module, an acceleration detection module, a mouth and nose respiration detection module, a signal self-encoding module, an LSTM featureextraction module, a wireless communication module, a LSTM neural network training device and a processing diagnosis device based on the LSTM neural network, a thermistor sensor is used for detectingmouth and nose breathing airflow, and a positive electrode and a negative electrode used for detecting respiratory and electrocardiogram signals of the chest of a human body. By using the system, hospitalization is not needed, the system is simple, and physiological and psychological burdens cannot be caused. The device can detect the respiratory states of different parts of a human body in multiple directions, achieves the purpose of sleep apnea syndrome classification diagnosis through an LSTM neural network classification algorithm, is simple to operate, and can be used at home.