Residual and bidirectional LSTM for epileptic seizure detection.

Electroencephalogram (EEG) plays a pivotal role in the detection and analysis of epileptic seizures, which affects over 70 million people in the world. Nonetheless, the visual interpretation of EEG signals for epilepsy detection is laborious and time-consuming. To tackle this open challenge, we introduce a straightforward yet efficient hybrid deep learning approach, named ResBiLSTM, for detecting epileptic seizures using EEG signals. Firstly, a one-dimensional residual neural network (ResNet) is tailored to adeptly extract the local spatial features of EEG signals. Subsequently, the acquired features are input into a bidirectional long short-term memory (BiLSTM) layer to model temporal dependencies. These output features are further processed through two fully connected layers to achieve the final epileptic seizure detection. The performance of ResBiLSTM is assessed on the epileptic seizure datasets provided by the University of Bonn and Temple University Hospital (TUH). The ResBiLSTM model achieves epileptic seizure detection accuracy rates of 98.88-100% in binary and ternary classifications on the Bonn dataset. Experimental outcomes for seizure recognition across seven epilepsy seizure types on the TUH seizure corpus (TUSZ) dataset indicate that the ResBiLSTM model attains a classification accuracy of 95.03% and a weighted F1 score of 95.03% with 10-fold cross-validation. These findings illustrate that ResBiLSTM outperforms several recent deep learning state-of-the-art approaches.

[Correlation of Heart Rate and Geographical Environmental Factors with Reference Values of Left Ventricular Ejection Fraction in Healthy Adult Males].

Objective To establish a model for obtaining the reference values of left ventricular ejection fraction(LVEF) in Chinese healthy adult males by exploring the relationships of these reference values with heart rate and geographical environment factors. Methods LVEF and heart rate reference values (X1) were collected from 3502 healthy adult males from 2006 to 2016. Correlation analysis and ridge regression were employed to extract dependent geographical environment factors and predict the LVEF reference values. The Kriging interpolation was applied to reveal the spatial distribution of the LVEF reference values. Results LVEF and heart rate (X1) were significantly correlated with five geographical environment factors. LVEF was negatively correlated with heart rate (X1),latitude (X3),and annual range of temperature (X9) and positively correlated with annual mean air temperature (X6),annual mean relative humidity (X7),and annual precipitation amount (X8). The reference values of LVEF had a negative correlation with heart rate. The ridge regression equation of LVEF reference values and geographical environment factors was as follows:Y=68.464-0.0949X3-0.0619X6-0.00128X7+0.00069X8-0.0199X9±3.329. The equation of LVEF reference values with heart rate and geographical environment factors was Y=75.923-0.1035X1-0.0958X3-0.0741X6+0.00094X7+0.00081X8-0.0211X9±3.288. Conclusion The LVEF reference values among Chinese healthy adult males decreased from south to north. They can be determined based on the regression models after the geographical factors of a certain region are obtained. The new model offers a geographic basis for the establishment of LVEF reference values.

[Geographical distribution of the Serum creatinine reference values of healthy adults].

OBJECTIVE: To explore the relationship between serum creatinine (Scr) reference values in healthy adults and geographic factors and provide evidence for establishing Scr reference values in different regions. METHODS: We collected 29 697 Scr reference values from healthy adults measured by 347 medical facilities from 23 provinces, 4 municipalities and 5 autonomous regions. We chose 23 geographical factors and analyzed their correlation with Scr reference values to identify the factors correlated significantly with Scr reference values. According to the Principal component analysis and Ridge regression analysis, two predictive models were constructed and the optimal model was chosen after comparison of the two model's fitting degree of predicted results and measured results. The distribution map of Scr reference values was drawn using the Kriging interpolation method. RESULTS: Seven geographic factors, including latitude, annual sunshine duration, annual average temperature, annual average relative humidity, annual precipitation, annual temperature range and topsoil (silt) cation exchange capacity were found to correlate significantly with Scr reference values. The overall distribution of Scr reference values featured a pattern that the values were high in the south and low in the north, varying consistently with the latitude change. CONCLUSION: The data of the geographic factors in a given region allows the prediction of the Scr values in healthy adults. Analysis of these geographical factors can facilitate the determination of the reference values specific to a region to improve the accuracy for clinical diagnoses.