Dual channel LSTM based multi-feature extraction in gait for diagnosis of Neurodegenerative diseases

The performance of gait disturbances differ in various Neurodegenerative diseases (NDs), which is an important basis for the diagnosis of NDs. In the diagnosis, doctors can judge disease state by observing patients' gait features without quantification, such a subjective diagnosis has been seen as a problem because diagnostic results may differ among doctors. Moreover, there are some irresistible factors such as fatigue may effects diagnostic procedure. To make use of these observations, we build an automatic deep model based on Long Short-Term Memory (LSTM) for the gait recognition problem. In our model, a dual channel LSTM model is designed to combine time series and force series recorded from NDs patients for whole gait understanding. Experimental results demonstrate that our proposed model improves gait recognition performance compared to baseline methods. We believe the quantitative evaluation provided by our method will assist clinical diagnosis of Neurodegenerative diseases.