In this paper, a CNN-Random Forest (CNN-RF) model is proposed and it can be used to automatically classify the skin diseases, on the basis of a dataset with 15,000 images, to generate four categories which will include acnes, hair loss, nail fungus and skin allergy. The discriminative features are obtained, in terms of overtuned convolutional neural network, and the classification is performed, on the basis of the Random Forest of the obtained features. The general accuracy of the model proposed is 96.08, the precision, recall and F1-score values are great (performance) across all classes. Competitive analysis demonstrates the performance and increased interpretability to the existing approaches. These results are a positive sign that the suggested system can be used as an efficient clinical decision-support system in dermatological diagnosis.

Globally, patients with diabetes, diabetic retinopathy, cancer, and heart disease are growing rapidly in developed and developing countries. As a result of these ailments, the rate of human mortality and vision loss has risen dramatically. The design and development of computer-based prediction systems may facilitate the appropriate treatment of these four illnesses by medical professionals. For the design of an efficient and fast prediction (or classification) system, it is necessary to use efficient feature selection techniques to reduce the complexity of the feature space. If there are n features, then there is a possibility that 2(n) subsets of features can be created, and testing all of these subsets of selected features would require a significant amount of time. The suggested technique is to investigate the application of ant-lion based optimization to choose a subset of features. The chosen characteristics are used to train and evaluate four classifiers (and their ensemble) based on machine learning. The study used over three public benchmark datasets and one privately composed dataset, each one was disease-specific. The performance of the recommended strategy was evaluated using five performance assessment measures. This adjustment significantly improves the outcome. The strategy may decrease the initial feature set by up to 50% without impacting performance (in terms of accuracy). We can get maximum accuracies of 84.44% for the heart disease dataset, 79.99% for the diabetes dataset, 98.52% for the diabetic retinopathy dataset, and 97.18% for the skin cancer dataset. This empirical research will help doctors and all people make better decisions by giving them a second opinion.

In the modern era, healthcare systems are increasingly moving towards digitalization and integration with cloud technologies, driven by the need to manage and analyse vast amounts of patient data effectively. E-healthcare systems struggle to integrate multisource data from EHRs, wearables, sensors, and imaging due to format, quality, and frequency variations. IoT, Digital Twin (DT), and Deep Transfer Learning (DTL) further complicate data harmonization, synchronization, and predictive accuracy for effective patient health management. The objectives are to develop a unified framework for integrating multisource data from EHRs, wearables, sensors, and imaging, leveraging IoT, DT, and DTL to enhance data harmonization, synchronization, and predictive accuracy for improved patient health management. Cloud-Edge Collaborative Filtering (CECF) optimizes e-healthcare by preprocessing data at the edge, filtering in the cloud, and enhancing decision-making and privacy. Adaptive Federated Transfer Learning (AFTL) enables decentralized, personalized model training across devices, enhancing privacy, accuracy, and adaptation in e-healthcare. Hopfield Neural Networks (HNN) in cloud-based DTs enhance pattern recognition, anomaly detection, and real-time health monitoring in e-healthcare. Findings show that the system demonstrates high predictive accuracy 92%, efficient TL 85%, fast data processing 2 seconds, reduced operational costs of 20%, and improved patient outcomes, including a 95% satisfaction rate and 15% readmission rate and implemented in Python software. Future scope includes enhancing personalization with AI-driven models, improving real-time decision-making, expanding interoperability across systems, integrating genomic data, ensuring advanced security, and enabling global collaboration for optimized patient care.

In the context of cloud computing, preserving the privacy of big data while also allowing for secure access control is a critical concern. With the increasing adoption of cloud technology, it is imperative to address the challenges associated with safeguarding sensitive data while enabling authorized access. This paper develops an efficient privacy-preserving security model that uses cryptographic techniques to protect sensitive data and ensure that only authorized individuals can access it. The research puts together a secure data authentication technique, named secured privacy protection access control (SecPPAccess), allowing secured communication in cloud computing. For the protection of privacy for sensitive data, the protected transferring of data is commenced among the elements, like a user, cloud server, registration authority, key generation center and data owner, by using many phases mainly the key generation phase, setup phase, server registration, user registration, data upload, data encryption, requester authentication, data access, and data download phase. Here, a method is designed newly for securing data privacy using various operations, like secret keys, hashing, encryption, etc. The study proves that the initiated SecPPAccess model achieves the highest rate of detection of 0.85, the lowest usage for memory of 0.505 MB, and less computation time of 51.50 s.

Cardiac arrhythmias are major global health concern and their early detection is critical for diagnosis. This study comprehensively evaluates the effectiveness of CNNs and LSTMs for the classification of cardiac arrhythmias, considering three PhysioNet datasets. ECG records are segmented to accommodate around ∼10s of ECG data. Followed by transformation to scalograms using DWT for training VGG-16; and WTS for feature extraction and dimensionality reduction for training LSTM network. VGG-16 achieved 96.44% test accuracy while LSTM achieved 92%. Results also highlight the effectiveness of VGG-16 for short-duration ECG analysis, while LSTM excels in long-term monitoring on edge devices for personalized healthcare.

The process of feature selection (FS) is vital aspect of machine learning (ML) model's performance enhancement where the objective is the selection of the most influential subset of features. This paper suggests the Gravitational search optimization algorithm (GSOA) technique for metaheuristic-based FS. Glaucoma disease is selected as the subject of investigation as this disease is spreading worldwide at a very fast pace; 111 million instances of glaucoma are expected by 2040, up from 64 million in 2015. It causes widespread vision impairment. Optic nerve fibres can be degraded and cannot be replaced later in this disease. As a starting point, the retinal fundus images of glaucoma infected persons and healthy persons are used, and 36 features were retrieved from these images of public benchmark datasets and private dataset. Six ML models are trained for classification on the basis of the GSOA's returned subset of features. The suggested FS technique enhances classification performance with selection of most influential features. The eight statistical performance evaluating parameters along with execution time are calculated. The training and testing have been performed using a split approach (70:30), 5-fold cross validation (CV), as well as 10-fold CV. The suggested approach achieved 95.36 % accuracy. Due to its auspicious performance, doctors might use the suggested method to receive a second opinion, which would also help overburdened skilled medical practitioners and save patients from vision loss.

In today's world, data leakage on computer systems, Internet of Things (IoT) devices, or mobile devices carriages a significant threat due to weaker encryption or communication techniques, resulting in the loss of data items. Identifying the leakage of sensitive data during data transmission requires an appropriate technique. In the IoT environment, default permissions granted to devices often lead to data leakage. This proposed method offers data leakage security based on data sensitivity. However, classifying sensitive data is challenging due to its large volume and data transformation. The proposed technique utilizes the minimum redundancy and maximum relevancy (mRMR) technique for feature selection. It accurately detects confidential data better than existing state-of-the-art methods and can identify rephrased confidential contents using filter-based features for sequential data leak deterrence. Specifically, data leakage measurement can be achieved using DBSCAN (Density-based Spatial Clustering of Applications with Noise) to average F-statistical values measured over individual time steps and to ensure continuity between leakage data points.

The primary cause of irreversible blindness due to glaucoma is a silent, progressive disease with no noticeable symptoms. This eye disease gradually and rapidly damages the optic nerve, resulting in visual field defects. Glaucoma can cause substantial vision loss if left untreated. Early detection and proper treatment help limit those severe consequences. The benefits of screening for glaucoma while reducing the workload on eye specialists outweigh the extra effort required for screening. With unmanned aircraft, self-driving cars, facial recognition, and language processing, artificial intelligence (AI) has altered our way of life. AI is capable of outperforming humans in tasks like image recognition. Data analysis and processing are critical, as the volume of image data generated by ophthalmic imaging centers continues to grow at a breakneck pace. Glaucoma has been predicted using OCT and fundus images of prospective patients and for this prediction; AI can be employed to help medical practitioners to come out of these problems. In this paper, a novel AI and Internet of Things (IoT) based predictive modeling is proposed in which a bio-inspired and artificial intelligence based computing approach is employed for classification and prediction of glaucoma disease from Optical coherence tomography (OCT) images through continuous monitoring. That ultimately results in an improvement in healthcare by providing necessary medical instructions. We are the frontrunners to present a unique IoT embedded with artificial intelligence that supports Glaucoma screening, an automated and timely system based on the fusion of machine learning and bio-inspired computing approaches, in the form of this study.150 OCT pictures were utilized in the experiment, which were derived from a mixture of MENDELEY and a private dataset by a renowned eye physicians. This work presents a solution to the question of how to diagnose this condition at an early stage utilizing 45 critical characteristics retrieved using the ORB feature extractor and custom algorithms. Our suggested model has four dimensions; originally, these 45 features were reduced to 20% (i.e., 9) utilizing a statistically based univariate selection procedure. Following that, a Genetic Algorithm (GA) is used to discover an optimum subset of characteristics. These optimized characteristics are routed sequentially to state-of-the-art machine learning models (K-Nearest Neighbor (KNN), XGBoost, Random Forest, and Support Vector Machine (SVM)) for classification. Additionally to the above, the technique is fully integrated into an IoT framework and can be accessed remotely to aid ophthalmologists in diagnosing and treating glaucoma. Additionally, the proposed model facilitates the collection of health data from patients through IoT devices. It is concluded that out of four possible sets of results, GA-KNN based combination input of 9 features, enhanced the computed results with 99% accuracy for glaucoma recognition. The accuracy is obtained through a fivefold cross-validation technique. Because the proposed system has a brilliant ability to differentiate between healthy and glaucomatous eyes, this study will help to achieve high standards of glaucoma identification.

Image splicing forgery is a prevalent form of digital image manipulation where various portions from one or multiple images are combined to create a deceptive image that appears genuine. Detecting image splicing forgery is crucial for verifying the authenticity of an image. Image splicing forgery detection has grown significantly in recent years, with numerous detection approaches proposed in the literature. This paper presents a comprehensive survey and classification of existing image splicing forgery detection approaches, focusing on 2014 to 2023. This study reviews 88 research papers on splicing in the context of image forgery detection. A generalized structure is introduced, outlining the typical stages involved in the detection process. The paper thoroughly reviews the literature, providing an overview of both hand-crafted and advanced detection approaches researchers propose. Benchmark datasets are identified, including their limitations. The objective is to provide a clear and comprehensive understanding of image splicing forgery detection for researchers and practitioners interested in this area. This survey is a valuable resource, offering insights into the field’s current state and highlighting areas for future research and development.

Medical records are transmitted between medical institutions using cloud-based Electronic health record (EHR) systems, which are intended to improve various medical services. Due to the potential of data breaches and the resultant loss of patient data, medical organizations find it challenging to employ cloud-based electronic medical record systems. EHR systems frequently necessitate high transmission costs, energy use, and time loss for physicians and patients. Furthermore, EHR security is a critical concern that jeopardizes patient privacy. Compared to a single system, cloud-based EHR solutions may bring extra security concerns as the system architecture gets more intricate. Access control strategies and the development of efficient security mechanisms for cloud-based EHR data are critical. For privacy reasons, the Dynamic constrained message authentication (DCMA) technique is used in the proposed system to encrypt the outsourced medical data by using symmetric key cryptography, which uses the Seagull optimization algorithm (SOA) to choose the best random keys for encryption and then resultant data is hashed using the SHA-256 technique. The results of the proposed model are evaluated using performance metrics, and the model attained a security of about 98.58%, which is proven to be superior because it adopts advanced random secret key generation, which adds more security to the system.

Cancer of the breasts is a prevalent and possibly fatal disease that causes abnormal development of cells in breast tissue. It is the most prevalent tumor in women globally, and it has various subtypes that respond differently to treatments. Early detection, such as mammograms, is critical to enhancing outcomes since it allows for prompt intervention. Treatment options may include chemotherapy, radiation therapy, surgery, and hormone therapy, either alone or in combination, depending on the features and stage of the cancer. Breast cancer has profound emotional and psychological effects on individuals and their families in addition to its physical effects Ongoing research, public awareness campaigns, and advances in personalized medicine all contribute to the collective efforts aimed at lowering the incidence of breast cancer, improving early detection, and improving the overall quality of life for those affected. By seamlessly integrating state-of-the-art deep learning models: pre-trained ResNet and U-Net, this study pioneers a transformative approach to breast cancer diagnosis. ResNet's expertise in hierarchical feature learning is combined with U-Net's segmentation prowess to focus on digital mammogram-based feature extraction and early-stage identification. The collaborative synergy provides a solid foundation for the accurate detection of breast abnormalities. A neural network is introduced to augment this process for classification, raising diagnostic capabilities to new heights. The combination of pre-trained ResNet and U-Net models creates a dynamic feature extraction pipeline for capturing intricate patterns and segmenting region-specific abnormalities. This collaborative methodology enables the model to detect subtle nuances indicative of early-stage breast cancer, thereby facilitating early detection and intervention. The seamless integration of these models addresses the complexities of mammographic data, providing a complete solution for accurate and nuanced breast cancer detection. A neural network is added to the diagnostic pipeline for more precise classification. The neural network refines the diagnostic process by analyzing the extracted features, reducing false positives, and increasing specificity. This multi-layered approach represents a significant step forward in breast cancer diagnosis, providing a comprehensive tool that integrates feature extraction, early-stage identification, and classification, with the potential to transform clinical practices and improve patient outcomes. Extensive validation and clinical testing validate this transformative model's efficacy and reliability in real-world healthcare scenarios. The responsible use of this transformative tool is supported by ethical considerations such as patient privacy safeguards and adherence to informed consent principles. This study, as a pioneering effort in breast cancer diagnosis, achieves outstanding performance with an accuracy of 99%, precision of 98.6%, recall of 99.01%, and specificity of 98.9%, showcasing superior metrics, lays the groundwork for future innovations, encouraging improved accuracy, personalized treatment strategies, and, ultimately, improved healthcare outcomes for patients.

The SARS-CoV-2 coronavirus strain’s introduction in December 2019 resulted in the development of the new coronavirus disease, COVID-19. Following its first appearance, the virus quickly spread throughout the world and is now considered to be a pandemic. There were 6,885,962 recorded deaths and 689,853,908 confirmed cases as of April 4, 2023. It is essential to put in place a thorough testing strategy in order to stop the disease from spreading. Nonetheless, a number of testing approaches are presently under consideration due to a restricted inventory and a limited supply of testing equipment. Recent expert analysis suggests that images from chest computed tomography (CT) scans could reveal important information about COVID-19, which is why we are using this modality as a focus of our investigation. Moreover, numerous recent studies have demonstrated that selecting the most informative features from the subject images improves the classification models’ efficiency and shortens the time needed for training and testing. All of this encourages us to present a study that suggests a novel, efficient, and fast feature selection system as the core of the proposed highly competent clinical decision support system for COVID-19 infection prediction. Using a publicly accessible CT image dataset, this four-phase system divides the images into two categories: "COVID-19-infected human" and "healthy human". Pre-processing is the first step in the study, after which features from different categories in the images are extracted in the next phase. In the third phase, the most influential features are then selected using three algorithms: the Teaching Learning-Based Optimization Algorithm (TLBO), the Cuckoo Search Optimization Algorithm (CSO), and a proposed hybrid of these two. To the best of the authors’ knowledge, these algorithms have rarely been applied to feature selection for COVID-19 infection prediction, which highlights the originality and inventiveness of the work. Following that, five machine learning (ML) classifiers that made use of the features selected during the feature selection stage are used to categorize the chest CT scans. Multiple tests, implementing the 70:30 approach, are then conducted for thorough investigation, and several performance-measuring metrics were computed during each test. A thorough investigation has been performed through multiple tests, and during each test, several performance-measuring metrics have been computed. Such in-depth investigations have rarely been published in state-of-the-art studies. With the suggested methodology, a noteworthy categorization accuracy of 97.94

Glaucoma is one of the leading causes of visual impairment worldwide. If diagnosed too late, the disease can irreversibly cause severe damage to the optic nerve, resulting in permanent loss of central vision and blindness. Therefore, early diagnosis of the disease is critical. Recent advancements in machine learning techniques have greatly aided ophthalmologists in timely and efficient diagnosis through the use of automated systems. Training the machine learning models with the most informative features can significantly enhance their performance. However, selecting the most informative feature subset is a real challenge because there are 2n potential feature subsets for a dataset with n features, and the conventional feature selection techniques are also not very efficient. Thus, extracting relevant features from medical images and selecting the most informative is a challenging task. Additionally, a considerable field of study has evolved around the discovery and selection of highly influential features (characteristics) from a large number of features. Through the inclusion of the most informative features, this method has the potential to improve machine learning classifiers by enhancing their classification performance, reducing training and testing time, and lowering system diagnostic costs by incorporating the most informative features. This work aims in the same direction to propose a unique, novel, and highly efficient feature selection (FS) approach using the Whale Optimization Algorithm (WOA), the Grey Wolf Optimization Algorithm (GWO), and a hybridized version of these two metaheuristics. To the best of our knowledge, the use of these two algorithms and their amalgamated version for FS in human disease prediction, particularly glaucoma prediction, has been rare in the past. The objective is to create a highly influential subset of characteristics using this approach. The suggested FS strategy seeks to maximize classification accuracy while reducing the total number of characteristics used. We evaluated the efficacy of the proposed approach in classifying eye-related glaucoma illnesses. In this study, we aim to assist professionals in identifying glaucoma by utilizing a proposed clinical decision support system that integrates image processing, soft-computing algorithms, and machine learning, and validates it on benchmark fundus images. Initially, we extract 65 features from the 646 retinal fundus images in the ORIGA benchmark dataset, from which a subset of features is created. For two-class classification, different machine learning classifiers receive the elected features. Employing 5-fold and 10-fold stratified cross-validation has enhanced the generalized performance of the proposed model. We assess performance using several well-established statistical criteria. The tests show that the suggested computer-aided diagnosis (CAD) model has an F1-score of 97.50

In order to locate the prominent objects which are featured among set of input photos, or to address the issue of co-saliency identification, a new approach is presented in this study. This research study intends to address the issue of co-saliency identification via early conscientious strategy, in contrast to the majority of prior approaches that call for correspondence matching. The proposed technique intends to detect co-saliency prior to the focused attentive response since it does not require to match the relationship between the two input images. The proposed approach also blocks the additional salient items detections, which only appears solo in single photos provided by the joint information of the image pair datasets. Finally, this research study demonstrates through various experimental results how well our algorithm locates the co-salient patterns in all paired input images.

The Cocktail party problem, which is tracing and identifying a specific speaker’s speech while numerous speakers communicate concurrently is one of the crucial problems still to be addressed for automated speech recognition (ASR) and speaker recognition. In this study, we attempt to thoroughly explore traditional methods for speech separation in a cocktail party environment and further analyze traditional single-channel methods for example source-driven methods like Computational Auditory Scene Analysis (CASA), data-driven methods like non-negative matrix factorization (NMF), model-driven methods, customary multi-channel methods such as beamforming, blind source separation for multi-channel and the newly developed deep learning approaches such as meta-learning based methods, self-supervised learning. This paper further accentuates numerous datasets and evaluation metrics in the domain of speech processing & brings out the comparison between traditional methods and methods based on deep learning for speech separation. This study provides a basic understanding and comprehensive knowledge of state-of-the-art researches in the area of speech separation and serves as a brief overview to new researchers.

In order to find similar salient objects within a collection of images, this study presents an innovative deep end-to-end co-saliency recognition method. The existing methods for characterizing co-saliency primarily depend upon manually created measurements. However, the subjectivity and lack of adaptability of these methodologies results in weak generalization. Additionally, most methods isolate the extraction of characteristics from individual and groups of photographs, ignoring the relation among both of these characteristics that may improve the efficiency of the model. By using a multiple-stage representation for obtaining features from a CNN with high spatial resolution, the suggested method addresses these above issues. This study exploits the learnable consistency using the improved Convolutional auto encoder. At last, final co-saliency maps are generated by fusing the intra-image contrasts and the inter-images stable feature sets. Results from experiments show that the proposed method outperforms with other approaches in terms of efficiency.