ABSTRACT
The whole world is completely upset because of the unexpected ejection of a lethal disease called Covid-19. Every single region is absolutely closed because of the effect of Covid. To prevent the unfold of this unwellness, everybody needs to maintain social distancing. Students are considered as the eventual fate of the country. To save the understudies from this infection the academic institute has begun internet educating and learning. Yet, giving information in online mode has become a testing task for understudies similarly as a tutor. Because of e-learning, customize learning has become vanish. To help intelligent instructing and learning systems an upgraded model is needed to boost the academic activities. This paper presents a style of projected model utilizing Reinforcement learning. The reinforcement learning (RL) approach provides effective pedagogical strategies for educating the learners with their interest in the subject. With the assistance of RL, the introduced model chooses the training difficulty level of scholars and recommends the student's understanding level to access the reading content. The proposed structure is planned in such a manner with the goal that the educator isn't needed to continually screen the understudy. Experimental results show that these approaches scale back the number of attentions needed from the teacher and enhance the training capability of understudy. The presented framework enhances personalized learning. © 2021 IEEE.
ABSTRACT
The medical diagnosis can be enhanced by intelligent and automatic diagnosis through advances in Information and Communication Technologies (ICT) during the Covid-19 pandemic. This chapter discusses the current scenario, fundamental concepts, and existing solutions for diagnosing corona based diseases and their limitations. The chapter presents a generic and hybrid intelligent architecture for disease diagnosis. The architecture considers CT scanned images along with other fuzzy parameters and classifies the images into various disease categories using a convolutional neural network. The fuzzy convolutional neural network has experimented on 100 CT scanned images of lungs with additional fuzzy symptoms to prove the architecture's utility. The working of the convolutional layer, pooling layer, fully connected layer, fuzzy membership functions, and training data sets used in the experiment are discussed in detail in this chapter. The results are analyzed and presented graphically with improvement in accuracy, sensitivity, and precision. The chapter concludes with applications of the architecture for other disease diagnoses using radiology images and also discusses limitations and future work enhancement. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.
ABSTRACT
Introduction: SARS-CoV-2 infection ranges from self-limiting viral illness to acute respiratory distress syndrome. Diffuse alveolar injury from SARS-CoV-2 increases risk of alveolar rupture. Pneumothorax is a rare complication that has been documented in the literature. Objective: The study aims to investigate the significance of pneumothorax as a complication in patients with previously healthy lungs with acute respiratory failure due to SARS-CoV-2 infection. Methods: This is a case-control study of adult patients without existing lung disease managed for acute respiratory failure who developed pneumothorax as a complication of the disease. Patients with iatrogenic or traumatic pneumothorax, history of chronic lung disease, or previous pneumothorax were excluded. To avoid sample bias from excluding possible false-negatives, the control group (documented SARSCoV- 2 negative) also included patients from the same time period without SARS-CoV-2 testing and a remote unexposed group from a year before. Chi-square analysis was used to determine the relationship between the development of pneumothorax and SARS-CoV-2 infection, with the null hypothesis being no difference in the frequency of pneumothorax among positive SARS-CoV-2 patients and negative SARS-CoV-2 patients. Results: One-hundred-and-thirty charts of patients with symptoms of SAR-CoV-2 were reviewed. Thirty-four patients were documented to be SARS-CoV-2 positive, twelve of which had pneumothorax as a complication of the disease, ninety-one percent of whom were mechanically-ventilated. The control group had ninety-six patients with documented SARS-CoV-2 negative tests, were untested, or part of the historical group. Six patients from the control group had pneumothorax as a complication, two of which were negative for SARS-CoV-2, and three were untested. A Chi-square analysis yielded an X2 statistic of 17.7549 with a p-value of 0.000025.Discussion: Pvalue rejects the null hypothesis in 0.1, 0.05, and 0.01 levels of significance which means that there is a statistically significant difference in the frequency of pneumothorax between the case and control groups. Calculations were done under the assumption that those with negative SARS-CoV-2 tests were indeed free of the virus. There is a probability of underestimation as the tests used may have not been 100-percent sensitive. The four cases of pneumothorax in the control group may have been false-negatives as these patients exhibited imaging findings suggestive of SARS-CoV-2 pneumonia and were likely infected with the virus as well.Conclusion: Pneumothorax is a significant complication in patients without existing lung disease who develop acute respiratory failure and SARS-CoV-2 infection. It should be anticipated and suspected when clinical deterioration occurs especially in mechanically-ventilated patients. .
ABSTRACT
This chapter discusses hybrid computational intelligence systems in various domains. The major techniques which are considered for the hybridization are the neural network, fuzzy logic, and genetic systems. Various neuro-fuzzy systems for course selection for students, web page classifications, finding a suitable profile for matrimonial applications and job, student aptitude testing, diagnosis, and emotional detection are presented in this chapter. The fuzzy convolutional neural network for COVID-19 analysis from various CT scanned lung images is also discussed in this chapter with necessary details such as neural network architecture, fuzzy membership functions, and training data sets. The chapter also discuss genetic fuzzy systems for fashion design and evolving neural network topologies with the required details. The genetic encoding, operations, neural network architecture, and fuzzy membership functions for these applications are discussed in detail. The other examples and applications demonstrated in the chapter include fuzzy collaborative filtering of movie recommendations, software quality evaluation using neuro-fuzzy-genetic hybridization, and introduction to the application of type-2 neuro-fuzzy system for the reliability of software products. For most of the applications included in this chapter, generic process flows and architectures of the system are illustrated graphically along with other details. These architecture are often having multiple layers and they are domain independent. The examples and applications discussed use these architectures to show the utilities of the architecture. At the end, the chapter enlists approximately 40 core and applied research and project possibilities in various areas. These applications include domain independent fuzzy activation function, fuzzy ontology as knowledge representation structure, idea generation framework, automatic knowledge discovery, eLearning system, automatic evolution of decision trees, healthcare advisory systems, product recommendation, and trust based network, consumer modeling, game playing, finance, marketing, etc. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.
ABSTRACT
The chapter provides examples and applications based on fuzzy sets and fuzzy logic-based theory. Initially, basic and fundamental examples of day to day life are demonstrated in this chapter with necessary design details and step by step calculations. The examples included here are fuzzification of irregular students considering fuzzy attendance, speed of a vehicle, job selection, fuzzy operations for almond sorting, and viral disease diagnosis such as the Covid-19. Numeric examples of fuzzification, defuzzification, operations on fuzzy sets, fuzzy relations are also included. Applications of fuzzy logic in fashion designing, software engineering, domestic appliances such as washing machines, share market analysis, and sensor control are also discussed in this chapter. Detailed discussion is presented on restaurant menu planner and customized representation of material to slow learners by giving complete systems architectures, design of fuzzy functions, and fuzzy rules. Traditional fuzzy logic, which is known as type-1 fuzzy logic, has got some limitations. To overcome the limitations, type-2 fuzzy logic is used. This chapter introduces and demonstrates an application of type-2 fuzzy logic along with its membership function. The fuzzy logic as a constituent of computational intelligence evolves continuously and observes possibilities of many innovative research opportunities. Besides detailed discussion on approximately 20 examples as mentioned above, in the end, the chapter enlists possible research ideas in the pure fuzzy logic-based system. There are possibilities of hybrid and applied research in the field of fuzzy logic too, which are enlisted at the end of the chapter. Approximately 40 core research ideas/projects and applications, which will be helpful for the learners, professionals, and researchers, are contributed to this chapter. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.