China
Africa
Explore chapters and articles related to this topic
Challenges in Designing Software Architectures for Web-Based Biomedical Signal Analysis
Published in Aboul Ella Hassanien, Nilanjan Dey, Surekha Borra, Medical Big Data and Internet of Medical Things, 2018
Alan Jovic, Kresimir Jozic, Davor Kukolja, Kresimir Friganovic, Mario Cifrek
We wrote some of the machine-learning algorithms from scratch, particularly those pertaining to feature selection (e.g. symmetrical uncertainty, Chi square, ReliefF), as we did not find any implementation under permissive licenses. For reporting the results of the analysis process (final statistics and evaluation measures of results), we opted to use the JasperReports (JR) library [103]. JR is licensed under LGPL, briefly meaning that in order to keep the possibility of commercializing our platform, we can only use its API, not change its source code. The use of a separate reporting library should be considered suitable for a web platform in the case of biomedical signal analysis applications, as many different types of statistical results may be obtained from the analysis scenario (e.g. for classification: class distribution, confusion matrix, total classification accuracy, sensitivity, specificity, F1 measure, etc.) [104]. Some of the results may be presented in a tabular form or as a list of evaluation measures, while others may use pie charts, histograms and so on.
En-Fuzzy-ClaF
Published in Neeraj Mohan, Surbhi Gupta, Chuan-Ming Liu, Society 5.0 and the Future of Emerging Computational Technologies, 2022
Sourabh Shastri, Sachin Kumar, Kuljeet Singh, Vibhakar Mansotra
In this section, the experiment focuses on evaluating the prediction performance of each single classifier with cross-validation five-fold and ten-fold. The four evaluation measures that have been used are: accuracy, f-measure, ROC and kappa statistics. The experimental results attained are shown in Table 8.3 and Figure 8.3. It is clear from Table 8.3 that the BayesNet classifier performed better than other classifiers, with an accuracy of 93.85% for both five-fold and ten-fold. But it is not too good to choose this model as a tool to identify COVID-19 detection. So, we go for ensemble learning also.
WEL-ODKC: weighted extreme learning optimal diagonal-kernels convolution model for accurate classification of skin lesions
Published in The Imaging Science Journal, 2023
V. Auxilia Osvin Nancy, P. Prabhavathy, Meenakshi S. Arya
Some standard performance evaluation measures are utilized here for assessing the proposed model’s performance in classifying skin lesions such as precision, accuracy, recall, F-score, and so on. True positives are the percentage of positive images that are accurately classified (). The positive images which are classified wrongly are called false negatives (). The negative images that are misclassified as positive are termed false positives () and negative images classified correctly as negatives are true negatives (). The positive image mainly represents the malignant skin cancer lesion that needs to be detected and in the negative image, the presence of the skin cancer lesion is absent. In the proposed model, the positive images are the ones that belong to the BCC, DF, MEL, and NEV classes. The negative images are the ones that belong to the benign category.
Integrating knowledge from DEX hierarchies into a logistic regression stacking model for predicting ski injuries
Published in Journal of Decision Systems, 2018
Boris Delibašić, Sandro Radovanović, Miloš Jovanović, Marko Bohanec, Milija Suknović
The experimental results are shown in Tables 1 and 2. We report AUC, Classification Accuracy (CA), F1 (harmonic mean between Precision and Recall), Precision and Recall, although in the paper from Bohanec and Delibašić (2015) only CA was reported due to DEX constraints for calculating the other evaluation measures. In both tables, LR (logistic regression) basic includes only the six basic attributes for prediction. LR DEX basic is a stacked model based on DEX hierarchy (Figure 1(a)). LR enhanced utilises all attributes (basic and enhanced from Table 1) for building the model. LR DEX enhanced is a stacked model based on the DEX hierarchy (Figure 1(b)).
An Efficient Hybrid Model for Acute Myeloid Leukaemia detection using Convolutional Bi-LSTM based Recurrent Neural Network
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
In this section, the simulation results for the AML detection are described using the available database, namely Munich AML Morphology Dataset and CPTAC- AML. This section explains the proposed detection of AML using the classification of AML cells from the group of images implemented in the MATLAB platform with Windows 8 OS followed by 4GB RAM. The evaluation measures, such as accuracy, precision, F1-score, sensitivity and specificity, were utilised for the evaluation of the proposed classifier. The proposed approach performance is compared with the several existing methods for the detection of AML.