Computational Intelligence in the Identification of COVID-19 Patients
Abdel-Badeeh M. Salem in Innovative Smart Healthcare and Bio-Medical Systems, 2020
Ozturka et al. [3] proposed DarkCovidNet models to provide accurate diagnostics for the classification approach when applied a COVID-19 X-ray image as data set. This data set comprised 43 female and 82 male COVID-19 positive patients. Neural networks (NN) approach was used for extracts features. DarkCovidNet models achieved 87.02% of accuracy. Mohamed et al. [4] analyzed seven deep learning models and studied the performance for each model by applying many of categories of classification such as binary and multiclass. The types of classification are support vector machine (SVM), artificial neural network (ANN), Naive Bayes and random forests. In machine learning approaches, NN model gives the highest detection rate when used (CSE-CIC-IDS2018) as data set. Data mining model used for the prediction of COVID-19-infected patients of South Korea was proposed by Muhammad et al. [5]. In their work, six algorithms (KNN, SVM, logistic regression, Naive Bayes, random forest, and DT) were applied directly on the COVID data set. The authors found that decision tree data mining algorithm is more efficient to predict the infected patients, with an accuracy of 99.85%. Wang et al. [6] used 1,065 computed tomography (CT) images of COVID-19 cases. In that study, accuracy, sensitivity, PPV.NPV, specificity, and area under curve (AUC) were used as markers of the classification process.
Big Data Analytics for COVID-19
Gitanjali Rahul Shinde, Asmita Balasaheb Kalamkar, Parikshit N. Mahalle, Nilanjan Dey in Data Analytics for Pandemics, 2020
Discussion: This technique is widely used in computing data analytics and mining and in statistical modeling. When we work on a huge dataset consisting of the details of underlined application and new observation or reading comes in, this technique is used to decide the category of this new observation. This technique requires trained (i.e. accurately collected observations) past data for classification. This technique is applicable to both structured as well as unstructured datasets. Classification can be binary classification, multi-class classification, or multi-label classification, and it is carried as per the underlined application. Initializing the required classifier, training it, predicting the target feature, and then evaluating this classifier for the required performance metrics are major steps in using the classification model. K-nearest neighbor and decision trees are some examples of classification techniques [30].
Combined models of artificial immune systems
Waldemar Wójcik, Andrzej Smolarz in Information Technology in Medical Diagnostics, 2017
Many discriminative methods, including Support Vector Machine, neural network and classifiers based on the artificial immune systems, are often most accurate and efficient when dealing with two classes only (they can deal with more classes, but usually at reduced accuracy and efficiency) (Ding 2004). For a large number of classes, higher-level multi-class methods are developed that utilise these two-class classification methods as the basic building blocks. To solve our problem, we used the strategy of one-versus-all based on the hybrid algorithm we developed the negative selection and artificial immune network. The simplest approach is to reduce the problem of classifying among K classes into K binary problems, where each problem discriminates a given class from the other K minus 1 classes. For this approach, we require N = K binary classifiers, where the k-th classifier is trained with positive examples belonging to class k and negative examples belonging to the other K minus 1 classes. When testing an unknown example, the classifier producing the maximum output is considered the winner, and this class label is assigned to that example. Rifkin and Klautau state that this approach, although simple, provides performance that is comparable to other more complicated approaches when the binary classifier is well-tuned (Rifkin & Klautau 2004).
Classification using semiparametric mixtures
Published in Journal of Applied Statistics, 2019
Classification is about assigning observations to known categories, based on the information provided by predictor variables. It is well-known to have a great many practical applications [7,8,22,26]. One can, for example, be interested in classifying tumors using gene expression data in cancer research [11], detecting spam emails [22], or assessing the accuracy of the labels for food samples in food authenticity studies [34]. A number of classification methods have been proposed in the literature and widely used in practice, with many described in the afore-mentioned books. Traditionally, the problem is known as discriminant analysis, which includes, e.g. the linear discriminant analysis [5,13] and quadratic discriminant analysis (e.g. [15,22]). The k-nearest-neighbor method, originally proposed in [14], labels a new observation with the class that is the most common among its k nearest observations. Classification trees were well studied and extended in [8], where tree-structured models are constructed and each leaf node is designated with a class label. Kernel density estimation can also be employed to estimate each class density, which is known as kernel discriminant analysis [39]. Support vector machines classify observations through decision functions and are among the most accurate classifiers in the literature [42].
Risks of cardiovascular toxicities associated with ALK tyrosine kinase inhibitors in patients with non-small-cell lung cancer: a meta-analysis of randomized control trials
Published in Expert Opinion on Drug Safety, 2023
Jin Zhao, Zhuo Ma, Hao Li, Dan Sun, Yi Hu, Chen Zhang, Yuhui Zhang
To further study the cardiovascular toxicities of ALK-TKIs, we classified cardiovascular adverse events as cardiac disorders, VTEs, and hypertension. The specific classifications are as follows, Cardiac disorders: myocardial ischemia; coronary artery disease; myocardial Infarction; bradycardia; atrial fibrillation; atrioventricular block; supraventricular tachycardia; atrial flutter; nodal rhythm; sinus node dysfunction; cardiac failure; left ventricular dysfunction; cardiac arrest; cardiac disorder; electrocardiogram T wave inversion; QT prolongation;VTEs: pulmonary embolism; deep venous thrombosis and other venous thrombosis events; also, thromboembolic event, venous thrombosis, thrombosis, and embolism were included;Hypertension.
Diurnal and seasonal changes in semen quality of men in subfertile partnerships
Published in Chronobiology International, 2018
Min Xie, Khyra Sarah Utzinger, Kerstin Blickenstorfer, Brigitte Leeners
The data were divided into four groups by the season in which the sample was collected, to compare the results of the semen quality. For the division into the four seasons, we used the astronomical system, as opposed to the meteorological system. This was based on various reasons. First, in the astronomical system, the seasons are divided according to the solstice and the equinox. The solstice describes the longest and the shortest day of the year and takes place on the 21st of June and the 21st or 22nd of December (Alexander et al. 2017). The equinox characterises the two days of the year where the duration of the day is equal to the duration of the night. These two days determine the change of the astronomical season from winter to spring and from summer to autumn. The vernal equinox takes place around the 21st of March and the autumnal equinox around the 23rd of September (Roenneberg and Aschoff 1990a). Since we suspected that the length of the day effected semen quality, this classification is the most sensible. Furthermore, using this definition allows comparison with other authors (Levine 1994). The following categories were used for data classification:
Related Knowledge Centers
- Cluster Analysis
- Cognition
- Cognitive Skill
- Prototype Theory
- Experience
- Subject & Object
- Idea
- Psychology
- Decision-Making
- Mental Representation