Computational Intelligence Techniques Research Articles

Preemptive Diagnosis of Alzheimer’s Disease in the Eastern Province of Saudi Arabia Using Computational Intelligence Techniques

Alzheimer's Disease (AD) is a silent disease that causes the brain cells to die progressively, influencing consciousness, behavior, planning ability, and language to name a few. AD increases exponentially with aging, where it doubles every 5-6 years, causing profound implications, such as swallowing difficulties and losing the ability to speak before death. According to the Ministry of Health in Saudi Arabia, AD patients will triple by 2060 to reach 14 million patients worldwide. The rapid rise of patients is caused by the silent progress of the disease, leading to late diagnosis as the symptoms will not be distinguished from normal aging affect. Moreover, with the current medical capabilities, it is impossible to confirm AD with 100% certainty via specific medical examinations. The literature review revealed that most recent publications used images to diagnose AD, which is insufficient for local hospitals with limited imaging capabilities. Other studies that used clinical and demographical data failed to achieve adequate results. Consequently, this study aims to preemptively predict AD in Saudi Arabia by employing machine learning (ML) techniques. The dataset was acquired from King Fahad Specialist Hospital (KFSH) in Dammam, Saudi Arabia, containing standard clinical tests for 152 patients. Four ML algorithms, namely, support vector machine (SVM), k-nearest neighbors (k-NN), Adaptive Boosting (AdaBoost), and eXtreme Gradient Boosting (XGBoost), were employed to preemptively diagnose the disease. The empirical results demonstrated the robustness of SVM in the pre-emptive diagnosis of AD with accuracy, precision, recall, and area under the receiver operating characteristics (AUROC) of 95.56%, 94.70%, 97.78%, and 0.97, respectively, with 13 features after applying the sequential forward feature selection technique. This model can assist the medical staff in controlling the progression of the disease at low costs.

RELATIONSHIP BETWEEN CBS QUALITY PARAMETERS FOR ASSESSMENT OF COMPUTATIONAL INTELLIGENCE TECHNIQUES

Software reliability plays a vital role in the emerging field of digitalization. Everyone wants cost and time-efficient software along with reliability which is achieved using CBS. In CBS, if the individual components are computed for a large or complicated system, then integration becomes complex which results in difficulty in predicting CBSR. To solve this problem several computational intelligence techniques such as SVM, ACO, PSO, ABC, GA, Neural network, are used but in our paper, we have focused on optimization techniques Fuzzy, ACO, ABC, PSO. These techniques help in estimating and predicting reliability models for CBS. Also, we have done, an assessment and comparative analysis based on a literature review of ABC, ACO, and PSO that have also been presented, for choosing suitable parameters for software reliability modeling.

A Novel IBA-DE Hybrid Approach for Modeling Sovereign Credit Ratings

Nowadays, the sovereign credit rating is not only an index of a country’s economic performance and political stability but also an overall indicator of development and growth, as well as the trust factor that is associated with the country. Due to its importance, the vast amount of available information, and the lack of a closed-form solution, prediction models based on machine learning (ML) and computation intelligence (CI) techniques are being increasingly used to complement traditional financial approaches. In this paper, we aim to introduce a novel ML-CI approach for sovereign credit rating prediction based on a differential evolution (DE) algorithm and interpolative Boolean algebra (IBA). In fact, the proposed approach is based on a pseudo-logical function in the IBA framework derived from the historical data of publicly available indicators using the DE algorithm. Such functions are easily interpreted and enable a subtle gradation among countries. It is shown that the IBA-DE approach outperforms back-propagation neural networks on the observed problem while also providing a deeper insight into each of the indicators used for prediction and its respective influence on the prediction rating on the other.

Role of Inflammation, Oxidative Stress, and Mitochondrial Changes in Premenstrual Psychosomatic Behavioral Symptoms with Anti-Inflammatory, Antioxidant Herbs, and Nutritional Supplements.

Premenstrual syndrome (PMS) significantly lowers the quality of life and impairs personal and social relationships in reproductive-age women. Some recommendations are that inappropriate oxidative stress and inflammatory response are involved in PMS. Various nutritional supplements and herbs showed neuro-psycho-pharmacological activity with antioxidant and anti-inflammatory properties. This study aims to determine the systematic review of randomized controlled trials (RCTs) of herbal medicine and nutritional supplements in PMS. We also comprehensively highlighted the role of oxidative stress, inflammation, and mitochondrial changes on PMS with the application of computational intelligence. We used PRISMA and research question-based techniques to collect the data for evaluation of our study on different databases such as Scopus, PubMed, and PROSPERO from 1990 to 2022. The methodological quality of the published study was assessed by the modified Jadad scale. In addition, we used network visualization and word cloud techniques to find the closest terms of the study based on previous publications. While we also used computational intelligence techniques to give the idea for the classification of experimental data from PMS. We found 25 randomized controlled studies with 1949 participants (mean ± SD: 77.96 ± 22.753) using the PRISMA technique, and all were high-quality studies. We also extracted the closest terms related to our study using network visualization techniques. This work has revealed the future direction and research gap on the role of oxidative stress and inflammation in PMS. In vitro and in vivo studies showed that bioactive molecules such as curcumin, allicin, anethole, thymoquinone, cyanidin 3-glucoside, gamma-linoleic acid, and various molecules not only have antioxidant and anti-inflammatory properties but also other various activities such as GABA-A receptor agonist, serotonergic, antidepressant, sedative, and analgesic. Traditional Unani Herbal medicine and nutritional supplements can effectively relieve PMS symptoms as they possess many bioactive molecules that are pharmacologically proven for the aforementioned properties. Hence, these biomolecules might influence a complex physical and psychological disease process like PMS. However, more rigorous research studies are recommended for in-depth knowledge of the efficacy of bioactive molecules on premenstrual syndrome in clinical trials.

Predicting the effects of selected reservoir petrophysical properties on bottomhole pressure via three computational intelligence techniques

This study investigates the effects of selected petrophysical properties on predicting flowing well bottomhole pressure. To efficiently situate the essence of this investigation, genetic, imperialist competitive and whale optimization algorithms were used in predicting the bottomhole pressure of a reservoir using production data and some selected petrophysical properties as independent input variables. A total of 15,633 data sets were collected from Volvo field in Norway, and after screening the data, a total of 9161 data sets were used to develop apt computational intelligence models. The data were randomly divided into three different groups: training, validation, and testing data. Two case scenarios were considered in this study. The first scenario involved the prediction of flowing bottomhole pressure using only eleven independent variables, while the second scenario bothered on the prediction of the same flowing bottomhole pressure using the same independent variables and two selected petrophysical properties (porosity and permeability). Each of the two scenarios involved as implied in the first scenario, the use of three (3) heuristic search optimizers to determine optimal model architectures. The optimizers were allowed to choose the optimal number of layers (between 1 and 10), the optimal number of nodal points (between 10 and 100) for each layer and the optimal learning rate required per task/operation. the results, showed that the models were able to learn the problems well with the learning rate fixed from 0.001 to 0.0001, although this became successively slower as the leaning rate decreased. With the chosen model configuration, the results suggest that a moderate learning rate of 0.0001 results in good model performance on the trained and tested data sets. Comparing the three heuristic search optimizers based on minimum MSE, RMSE, MAE and highest coefficient of determination (R2) for the actual and predicted values, shows that the imperialist competitive algorithm optimizer predicted the flowing bottomhole pressure most accurately relative to the genetic and whale optimization algorithm optimizers.

EEG Feature Extraction Using Evolutionary Algorithms for Brain-Computer Interface Development.

Brain–computer interfaces are systems capable of mapping brain activity to specific commands, which enables to remotely automate different types of processes in hardware devices or software applications. However, the development of brain–computer interfaces has been limited by several factors that affect their performance, such as the characterization of events in brain signals and the excessive processing load generated by the high volume of data. In this paper, we propose a method based on computational intelligence techniques to handle these problems, turning them into a single optimization problem. An artificial neural network is used as a classifier for event detection, along with an evolutionary algorithm to find the optimal subset of electrodes and data points that better represents the target event. The obtained results indicate our approach is a competitive and viable alternative for feature extraction in electroencephalograms, leading to high accuracy values and allowing the reduction of a significant amount of data.

Integration of Lineal Geostatistical Analysis and Computational Intelligence to Evaluate the Batch Grinding Kinetics

The kinetic characterization of the grinding process has always faced a special challenge due to the constant fluctuations of its parameters. The weight percentage of each size (WPES) should be mentioned. There are particular considerations for WPESs, because their tendencies are not monotonic. The objective of this work is to provide a methodology and model that will allow us to better understand the kinetics of grinding through the analysis of the Response Surface (RS), using geostatistical (data reconstruction) and computational intelligence (meta-model) techniques. Six experimental cases were studied and trends were evaluated/adjusted with multiple parameters, including an identity plot adjusted to 0.75–0.90, a standardized error histogram with a mean of −0.01 to −0.05 and a standard deviation of 0.63–1.2, a standardized error based on an estimated value of −0.09 to −0.02, a meta-model adjusted to between 92 and 99%, and finally, using the coefficient of variation, which classifies the information (stable/unstable). In conclusion, it was feasible to obtain the results of the WPES from RS, and it was possible to visualize the areas of greatest fluctuation, trend changes, error adjustments, and data scarcity without the need for specific experimental techniques, a coefficient analysis of the fracturing or the use of differential equation systems.

Permeability extraction from multiple well logs using particle swarm optimization based factor analysis

In this paper, we present an innovative factor analysis algorithm for hydrocarbon exploration to estimate the intrinsic permeability of reservoir rocks from well logs. Unlike conventional evaluation methods that employ a single or a limited number of data types, we process simultaneously all available data to derive the first statistical factor and relate it to permeability by regression analysis. For solving the problem of factor analysis, we introduce an improved particle swarm optimization method, which searches for the global minimum of the distance between the observed and calculated data and gives a quick estimation for the factor scores. The learning factors of the intelligent computational technique such as the cognitive and social constants are specified as hyperparameters and calculated by using simulated annealing algorithm as heuristic hyperparameter estimator. Instead of the arbitrary fixation of these hyperparameters, we refine them in an iterative process to give reliable estimation both for the statistical factors and formation permeability. The estimated learning parameters are consistent with literature recommendations. We demonstrate the feasibility of the proposed well-log analysis method by a Hungarian oilfield study involving open-hole wireline logs and core data. We determine the spatial distribution of permeability both along a borehole and between more wells using the factor analysis approach, which serves as efficient and reliable multivariate statistical tool for advanced formation evaluation and reservoir modeling.

LSTM-Based Prediction of Mediterranean Vegetation Dynamics Using NDVI Time-Series Data

Vegetation index time-series analysis of multitemporal satellite data is widely used to study vegetation dynamics in the present climate change era. This paper proposes a systematic methodology to predict the Normalized Difference Vegetation Index (NDVI) using time-series data extracted from the Moderate Resolution Imaging Spectroradiometer (MODIS). The key idea is to obtain accurate NDVI predictions by combining the merits of two effective computational intelligence techniques; namely, fuzzy clustering and long short-term memory (LSTM) neural networks under the framework of dynamic time warping (DTW) similarity measure. The study area is the Lesvos Island, located in the Aegean Sea, Greece, which is an insular environment in the Mediterranean coastal region. The algorithmic steps and the main contributions of the current work are described as follows. (1) A data reduction mechanism was applied to obtain a set of representative time series. (2) Since DTW is a similarity measure and not a distance, a multidimensional scaling approach was applied to transform the representative time series into points in a low-dimensional space, thus enabling the use of the Euclidean distance. (3) An efficient optimal fuzzy clustering scheme was implemented to obtain the optimal number of clusters that better described the underline distribution of the low-dimensional points. (4) The center of each cluster was mapped into time series, which were the mean of all representative time series that corresponded to the points belonging to that cluster. (5) Finally, the time series obtained in the last step were further processed in terms of LSTM neural networks. In particular, development and evaluation of the LSTM models was carried out considering a one-year period, i.e., 12 monthly time steps. The results indicate that the method identified unique time-series patterns of NDVI among different CORINE land-use/land-cover (LULC) types. The LSTM networks predicted the NDVI with root mean squared error (RMSE) ranging from 0.017 to 0.079. For the validation year of 2020, the difference between forecasted and actual NDVI was less than 0.1 in most of the study area. This study indicates that the synergy of the optimal fuzzy clustering based on DTW similarity of NDVI time-series data and the use of LSTM networks with clustered data can provide useful results for monitoring vegetation dynamics in fragmented Mediterranean ecosystems.

Prediction of compressive strength in plain and blended cement concretes using a hybrid artificial intelligence model

Concrete is a complex engineering material with different ingredients and design variables. Therefore, an accurate prediction of concrete compressive strength (CCS) should consider its nonlinearity. Data-driven machine learning approaches can solve highly non-linear problems. The objective of this paper is to utilize computational intelligence techniques, especially the support vector machine (SVM) and the genetic algorithm (GA) as the hybridized model (SVM-GA) to reliably predict concrete compressive strength based on a set of independent variables such as Ordinary Portland Cement (OPC), Silica Fume (SF), Fly As (FA), Cement Content, Water Content. Such an approach provides a fast and simple way to optimize parameter settings to obtain an optimum laboratory condition of the CCS process. The study shows that the concrete compression strength increases with increasing silica fume (SF), cement content and curing time, and a higher correlation coefficient in SF was observed compared to cement content. In addition, the correlation coefficients for Fly Ash (FA), water and aggregates were found to be −0.38, −0.28 and −0.28, respectively, which has indicated a significant decrease in concrete compressive strength with these parameters and that the addition of FA is more significant in decreasing the compressive strength of concrete than water content and both coarse and fine aggregates and this can be explained by the low pozzolanic reactivity of FA in the early stages of concrete hydration. The experimental results demonstrated that the proposed model has promising potential with correlations of R2 = 0.99 achieved during the testing stages of the prediction of concrete compression strength. Furthermore, the experimental results demonstrate that the predicted values obtained from the model (SVM-GA) correspond closely to the experimental results with the root mean square error (RMSE)= 0.002.

Comparative analysis of different computational intelligence techniques for maximum power point tracking of PV systems

Comparative analysis of different computational intelligence techniques for maximum power point tracking of PV systems

Intelligent Decision Support and Fusion Models for Fault Detection and Location in Power Grids

Computational intelligence-based diagnostic frameworks have emerged as rapidly evolving but highly efficient approaches for diagnosing faults in power grids. This work aims to build a diagnostic framework by resorting to computational intelligence techniques to improve decision-making and diagnostic accuracy. This diagnostic framework has three modules for signal processing, fault detection, and location. The signal-processing module uses the variational mode decomposition technique to extract informative time-frequency features from the voltage and frequency signals. Voltage features are then fed into the fault detection module to train a set of modular support vector machines that are used for monitoring the binary state of each node in the power grid. Once a faulty state on a node is detected, it activates the third module for identifying fault location. This module benefits from a novel zSlices-based general type-2 fuzzy fusion model for the sake of identifying the fault type as well as mitigating the false alarm rate. The exact location of the fault is then determined through a fuzzy decision support system that is equipped with a recommendation mechanism for the sake of consensus reaching. Various scenarios are simulated on the IEEE 39-bus system and on an experimental setup of a Three-Bus Two-Line transmission system, where the attained results verify the applicability, efficiency, and robustness of the proposed framework.

Swarm Intelligence with Adaptive Neuro-Fuzzy Inference System-Based Routing Protocol for Clustered Wireless Sensor Networks.

Wireless sensor network (WSN) comprises numerous compact-sized sensor nodes which are linked to one another. Lifetime maximization of WSN is considered a challenging problem in the design of WSN since its energy-limited capacity of the inbuilt batteries exists in the sensor nodes. Earlier works have focused on the design of clustering and routing techniques to accomplish energy efficiency and thereby result in an increased lifetime of the network. The multihop route selection process can be treated as an NP-hard problem and can be solved by the use of computational intelligence techniques such as fuzzy logic and swarm intelligence (SI) algorithms. With this motivation, this article aims to focus on the design of swarm intelligence with an adaptive neuro-fuzzy inference system-based routing (SI-ANFISR) protocol for clustered WSN. The proposed SI-ANFISR technique aims to determine the cluster heads (CHs) and optimal routes for multihop communication in the network. To accomplish this, the SI-ANFISR technique primarily employs a weighted clustering algorithm to elect CHs and construct clusters. Besides, the SI-ANFISR technique involves the design of an ANFIS model for the selection process, which make use of three input parameters, namely, residual energy, node degree, and node history. In order to optimally adjust the membership function (MF) of the ANFIS model, the squirrel search algorithm (SSA) is utilized. None of the earlier works have used ANFIS with SSA for the routing process. The design of SSA to tune the MFs of the ANFIS model for optimal routing process in WSN shows the novelty of the study. The experimental validation of the SI-ANFISR technique takes place, and the results are inspected under different aspects. The simulation results highlighted the significant performance of the SI-ANFISR technique compared to the recent techniques with a maximum throughput of 43838 kbps and residual energy of 0.4800J, respectively.

Computational intelligence based sustainable computing with classification model for big data visualization on map reduce environment

In recent years, the researchers have perceived the modifications or transformations motivated by the presence of big data on the definition, complexity, and future direction of the real world optimization problems. Big Data visualization is mainly based on the efficient computer system for ingesting actual data and producing graphical representation for understanding large quantity of data in a fraction of seconds. At the same time, clustering is an effective data mining tool used to analyze big data and computational intelligence (CI) techniques can be employed to solve big data classification process. In this aspect, this study develops a novel Computational Intelligence based Clustering with Classification Model for Big Data Visualization on Map Reduce Environment, named CICC-BDVMR technique. The proposed CICC-BDVMR technique intends to perform effective BDV using the clustering and data classification processes on the Map Reduce environment. For clustering process, a grasshopper optimization algorithm (GOA) with kernelized fuzzy c-means (KFCM) technique is used to cluster the big data and the GOA is mainly utilized to determine the initial cluster centers of the KFCM technique. GOA is a recently proposed metaheuristic algorithm inspired by the swarming behaviour of grasshoppers. This algorithm has been shown to be efficient in tackling global unconstrained and constrained optimization problems. Based on the modified GOA, an effective kernel extreme learning machine model for financial stress prediction was created. Besides, big data classification process takes place using the Ridge Regression (RR) and the parameter optimization of the RR model is carried out via the Red Colobuses Monkey (RCM) algorithm. The design of GOA and RCM algorithms for parameter optimization processes for big data classification shows the novelty of the study. A wide ranging simulation analysis is carried out using benchmark big datasets and the comparative results reported the enhanced outcomes of the CICC-BDVMR technique over the recent state of art approaches. The broad comparison research illustrates the CICC-BDVMR approach’s promising performance against contemporary state-of-the-art techniques. As a result, the CICC-BDVMR technique has been demonstrated to be an effective technique for visualising and classifying large amounts of data.

Structural damage identification employing hybrid intelligence using artificial neural networks and vibration-based methods

The basis of vibration-based damage detection method is that when there are changes in physical properties of a structure, there will also be changes in vibration characteristics such as natural frequencies, mode shapes and damping ratios. Artificial neural network has become one of the most powerful approaches capable of pattern recognition, classification, and nonlinear modelling employing computational intelligence techniques to tackle damage detection as a complex problem. In this paper, ensemble neural networks based damage identification techniques were developed and applied for damage localization and severity identification in I-beam structures using dynamic parameters. Experimental modal analysis and numerical simulations of I-beam with triple-point damage cases were carried out to generate the natural frequencies and mode shapes of structures. In the procedure of damage identification, five different neural networks corresponding to mode 1 to mode 5 were constructed, and then an approach based on a neural network ensemble was proposed to combine the outcomes of the individual neural networks into a single network. The ensemble neural network has the superiorities of all the individual networks from different vibrational modes. The ensemble network produced better damage identification outcomes than the individual networks and the results revealed the ability of the ensemble neural networks to identify damage.

Application of Visual Sensing Techniques in Computational Intelligence for Risk Assessment of Sports Injuries in Colleges.

Injury prediction is one of the most challenging issues in sports and is a key component of injury prevention, since successful injury prediction forms the basis for effective preventive measures. In this study, an analysis was made on the risk of physical injuries to college students to guarantee the physical safety of students in sports and improve the quality of physical education. Then, a study was carried out on the occurrences of physical injury risks through visual sensing techniques, and an investigation was conducted into the characteristics of physical injury risks in colleges. Next, the student's body shape and physical characteristics are computed using visual sensing techniques, and the risk of sports injuries is evaluated. The results show that the proposed image recognition and computation methods can accurately identify the sports injuries of college students. Furthermore, it can effectively analyze the factors affecting the risk of sports injuries, and the error of the proposed technique remains between −3 and 2. In addition, it can accurately locate the occurrence of sports injury risks and reduce the impact of those risks in time. This work provides technical support for the reduction of sports injury risks and contributes to the improvement of physical education teaching quality in colleges.

Stock Portfolio Management in the Presence of Downtrends Using Computational Intelligence

Stock portfolio management consists of defining how some investment resources should be allocated to a set of stocks. It is an important component in the functioning of modern societies throughout the world. However, it faces important theoretical and practical challenges. The contribution of this work is two-fold: first, to describe an approach that comprehensively addresses the main activities carried out by practitioners during portfolio management (price forecasting, stock selection and portfolio optimization) and, second, to consider uptrends and downtrends in prices. Both aspects are relevant for practitioners but, to the best of our knowledge, the literature does not have an approach addressing them together. We propose to do it by exploiting various computational intelligence techniques. The assessment of the proposal shows that further improvements to the procedure are obtained when considering downtrends and that the procedure allows obtaining portfolios with better returns than those produced by the considered benchmarks. These results indicate that practitioners should consider the proposed procedure as a complement to their current methodologies in managing stock portfolios.

X-ware: a proof of concept malware utilizing artificial intelligence

<p>Recent years have witnessed a dramatic growth in utilizing computational intelligence techniques for various domains. Coherently, malicious actors are expected to utilize these techniques against current security solutions. Despite the importance of these new potential threats, there remains a paucity of evidence on leveraging these research literature techniques. This article investigates the possibility of combining artificial neural networks and swarm intelligence to generate a new type of malware. We successfully created a proof of concept malware named X-ware, which we tested against the Windows-based systems. Developing this proof of concept may allow us to identify this potential threat’s characteristics for developing mitigation methods in the future. Furthermore, a method for recording the virus’s behavior and propagation throughout a file system is presented. The proposed virus prototype acts as a swarm system with a neural network-integrated for operations. The virus’s behavioral data is recorded and shown under a complex network format to describe the behavior and communication of the swarm. This paper has demonstrated that malware strengthened with computational intelligence is a credible threat. We envisage that our study can be utilized to assist current and future security researchers to help in implementing more effective countermeasures</p>

A neural network based heading and position control system of a ship

Heading and position control system of ships has remained a challenging control problem. It is a nonlinear multiple input multiple output system. Moreover, the dynamics of the system vary with operating as well as environmental conditions. Conventionally, simple Proportional Integral Derivative controller is used which has well known limitations. Other conventional control techniques have also been investigated but they require an accurate mathematical model of a ship. Unfortunately, accuracy of mathematical models is very difficult to achieve. During the past few decades computational intelligence techniques such as artificial neural networks have been very successful when an accurate mathematical model is not available. Therefore, in this paper, an artificial neural network controller is proposed for heading and position control system. For simulation purposes, a mathematical model with four effective thrusters have been chosen to test the performance of the proposed controller. The final closed loop system has been analysed and tested through simulation studies. The results are very encouraging.

Exergetic optimization of simple and finned solar air collectors for humid subtropical regions.

In this paper, an exergetic optimization of simple and finned flat plate solar air collectors is established to determine the optimal design and operational parameters for humid subtropical climatic conditions. Solar air collectors are commonly used for space heating, irrigation, greenhouses, grain drying, hot air generation, etc. A detailed solar energy radiation model is developed for solar energy conversion. A comprehensive optical, energy, and exergy analysis is carried out for evaluating the performance, energy and exergetic efficiency for simple and finned flat plate solar air collectors under humid subtropical climatic conditions. A simulation program is developed for solar energy model, energy and exergetic calculations. The following geometric and operating parameters are considered as decision variables: absorber plate area, dimensions of simple and finned solar collectors, fluid inlet and outlet temperatures, average velocity, overall loss coefficient, glass cover temperatures, plate temperature, and useful heat gain. The proposed model is optimized using computational intelligence technique (single phase multi-group teaching learning optimization) for maximum mean exergy efficiency in simple and finned solar air collectors for humid subtropical regions. The maximum exergy efficiency is achieved in the case of finned solar air collectors with a yearly optimal average exergy efficiency of 6.10% for a collector area of 5 m2 and a yearly average heat flux of 601W/m2. Thus, beneficial applications of exergetic optimization in design and operation of solar collectors for humid subtropical climatic conditions based on maximum exergy efficiency according to the optimized parameters and benefits of this approach for such systems have been highlighted.