Epilepsy is one of the most common neurological disorders generally characterized by sudden and recurrent unprovoked seizures, which is commonly diagnosed through visual inspection of electroencephalography (EEG) signals. However, since EEG signals are highly complex, nonlinear and nonstationary in nature, both visual inspection and the existing computed aid detection approaches fail to capture the intrinsic dynamics of seizure events, leading to unsatisfactory detection performance. Therefore, an accurate computer-aided automatic epileptic seizure diagnosis system is valuable and urgently needed in clinical application. In this study, we propose an automatic classification method using nonlinear dynamics features and nonlinear classifiers for determining whether EEG signals are in interictal or ictal activities, respectively. The main novelty of this paper is that investigating the performance of discernable entropy-based nonlinear dynamics features used for automatic diagnosis of epileptic seizure. First, discrete wavelet transformation is applied to obtain de-noised and sub-band signals. Second, the nonlinear dynamics features are extracted based on information entropy theory. Subsequently, one-way analysis of variance and forward sequential feature selection technique are employed to identify the most explainable features for seizure detection. Then, the selected features are fed to six different classifiers, namely, least square-support vector machine (LS-SVM), K-nearest neighbors, logistic regression, linear discriminant analysis, Naive Bayes classifier, and random forest, for automatic diagnosis of epileptic seizure. Finally, a public EEG dataset from Bonn University, Germany is utilized to investigate the performance of the proposed method. Experimental results show that LS-SVM yields the optimal performance with an accuracy of 99.50INDEX TER.

Nowadays, air pollution is a severe environmental problem in China. To investigate the effects of ambient air pollution on health, a time series analysis of daily outpatient and inpatient visits in 2015 were conducted in Shenzhen (China). Generalized additive model was employed to analyze associations between six air pollutants (namely SO₂, CO, NO₂, O₃, PM₁₀, and PM_2.5) and daily outpatient and inpatient visits after adjusting confounding meteorological factors, time and day of the week effects. Significant associations between air pollutants and two types of hospital visits were observed. The estimated increase in overall outpatient visits associated with each 10 µg/m³ increase in air pollutant concentration ranged from 0.48% (O₃ at lag 2) to 11.48% (SO₂ with 2-day moving average); for overall inpatient visits ranged from 0.73% (O₃ at lag 7) to 17.13% (SO₂ with 8-day moving average). Our results also suggested a heterogeneity of the health effects across different outcomes and in different populations. The findings in present study indicate that even in Shenzhen, a less polluted area in China, significant associations exist between air pollution and daily number of overall outpatient and inpatient visits.

The individual physiological difference has been recognized as one of the major problems while assessing subjects using multiple physiological data modeling and fusion techniques. To address this issue, we propose a one-step tensor-based modeling procedure to fuse the subject-variant information and multi-channel physiological data. Specifically, we consider the information similarity of the information matrixes from the tensor decomposition while introducing the subject-variant information, and form a tensor-based optimization problem to achieve the goal of physiological status assessment. To well solve this problem, a tailored alternating direction method of multipliers (ADMM) embedded block coordinate descent (BCD) algorithm has been proposed. Four real-case datasets from different scenarios have been employed to validate our proposed approach, and the performance indicates the superiority compared to several existing methods. Note to Practitioners—The proposed method aims to assess the physiological status by fusing multi-channel physiological data and subject-variant information. To better implement this method, three things are noteworthy. First, the dataset used in the proposed method should contain aligned multi-channel physiological data and subject-variant data, that is, the multi-channel physiological data and the subject-variant data should be correspondingly related. Second, the size of physiological data segments should be moderate due to the limited number of physiological data channels. Third, the initial value of the rank of CANDECOMP/PARAFAC (CP) tensor decomposition, $\textbf{\textit{k}}$ , should be carefully chosen with the contextual knowledge. The value of $\textbf{\textit{k}}$ is suggested not higher than the number of channels.

Foodborne disease is a growing public health problem worldwide and imposes a considerable economic burden on hospitals and other healthcare costs. Thus, accurately predicting the propagation of foodborne disease is crucial in preventing foodborne disease outbreaks. Few studies have investigated the dependencies between environmental variables and foodborne disease activity. This study develops a regularization-based eXtreme gradient boosting approach for foodborne disease trend forecasting considering environmental effects to capture dependencies hidden in foodborne disease time series. A real case in Shanghai, China was studied to validate our proposed model along with comparisons to traditional and benchmark algorithms for foodborne disease prediction. Results show that the foodborne disease prediction approach we propose achieves slightly superior performance in terms of one-day-ahead prediction of foodborne disease, and presents more robust prediction for 2-7 days ahead prediction.