METHODS: The derivation cohort included 90 Malaysian GBS patients with two sets of NCS performed early (1-20days) and late (3-8 weeks). Potential predictors of AIDP were considered in univariate and multivariate logistic regression models to develop a predictive model. The model was externally validated in 102 Japanese GBS patients.

RESULTS: Median motor conduction velocity (MCV), ulnar distal motor latency (DML) and abnormal ulnar/normal sural pattern were independently associated with AIDP at both timepoints (median MCV: p = 0.038, p = 0.014; ulnar DML: p = 0.002, p = 0.003; sural sparing: p = 0.033, p = 0.009). There was good discrimination of AIDP (area under the curve (AUC) 0.86-0.89) and this was valid in the validation cohort (AUC 0.74-0.94). Scores ranged from 0 to 6, and corresponded to AIDP probabilities of 15-98% at early NCS and 6-100% at late NCS.

CONCLUSION: The probabilities of AIDP could be reliably predicted based on median MCV, ulnar DML and ulnar/sural sparing pattern that were determined at early and late stages of GBS.

METHODS: A convolutional auto-encoder (CAE) based nonlinear compression structure is implemented to reduce the signal size of arrhythmic beats. Long-short term memory (LSTM) classifiers are employed to automatically recognize arrhythmias using ECG features, which are deeply coded with the CAE network.

RESULTS: Based upon the coded ECG signals, both storage requirement and classification time were considerably reduced. In experimental studies conducted with the MIT-BIH arrhythmia database, ECG signals were compressed by an average 0.70% percentage root mean square difference (PRD) rate, and an accuracy of over 99.0% was observed.