Multi-class medical image classification using DL continues to face major challenges, including managing multi-modal data, adapting to new tasks, handling distributed datasets, and operating under limited computational resources. Existing approaches fail to address these issues simultaneously, restricting the clinical scalability of AI in healthcare. To overcome these limitations, this paper introduces BigOrthoATD.Net, a unified, serverless, and decentralized learning framework that redefines scalability, adaptability, and efficiency in orthopedic image analysis. Designed to operate across distributed clinical nodes, BigOrthoATD.Net enables privacy-preserving knowledge fusion and multimodal integration across X-ray and CT imaging modalities. The framework supports progressive scalability for new tasks and institutions, achieving continual learning without retraining or performance degradation. Comprehensive experiments conducted across 13 simulated decentralized nodes and 50 orthopedic classes demonstrated that BigOrthoATD.Net achieved a state-of-the-art accuracy of 97.0%, outperforming swarm learning (70.8%) and centralized learning (84.8%), while federated learning failed to converge beyond moderate scale under identical resource-constrained conditions. BigOrthoATD.Net establishes a new benchmark for decentralized medical imaging by surpassing both centralized and decentralized frameworks in accuracy, scalability, and class diversity, while operating efficiently in low-resourced settings.

The existing oil reservoir demonstrated suboptimal inter-well connectivity, leading to irregular depletion and reduced overall production efficiency. This article demonstrates the Attention-Guided Fusion Model for Injector–Producer Connectivity Estimation (AGFM). The model has an attention mechanism in the first path, pulling discernment from the relationships between injectors and producers through the training phase, extracting the attention weight. This attention weight is then devoted to the second path, utilising a Long-Short-Term Memory (LSTM)-based architecture. The first path is only to the training stage. In contrast, the second path is used during training and testing, improving the ability of the model to find a more significant representation of the data. This makes the model robust enough to predict reservoir performance and interconnectivity, giving valuable insights to optimise field operations. The AGFM undergoes an evaluation with two different injection liquids (carbon dioxide ( C O 2 ) and water) in three scenarios: all water and all CO2 alternating between water and CO2 as a flooding liquid. The evaluation emphasises the efficacy of the model in all scenarios, making it a practical tool for estimating reservoir connectivity and enhancing oil recovery strategies. The water alternating gas (WAG) process performed high accuracy rates, with 82.1% for oil production, 86.8% for water production, and 86.9% for gas production. Our proposed method consistently demonstrates superior performance through comprehensive experimentation and rigorous analysis compared to existing approaches. The results reveal spatial interwell connectivity, confirming the efficacy and potential of our method as a more effective solution for reservoir recovery. • Present novel Attention-based Graph Fusion Model for injector–producer connectivity. • Utilisation of an attention mechanism to allocate weights to crucial connectivity aspects. • Development of an IPM-based guidance method for accurate identification. • Advanced techniques to enhance the precision of connectivity predictions.

ABSTRACT This study introduces a new multi‐criteria decision‐making (MCDM) framework to evaluate trauma injury detection models in intensive care units (ICUs). This research addresses the challenges associated with diverse machine learning (ML) models, inconsistencies, conflicting priorities, and the importance of metrics. The developed methodology consists of three phases: dataset identification and pre‐processing, hybrid model development, and an evaluation/benchmarking framework. Through meticulous pre‐processing, the dataset is tailored to focus on adult trauma patients. Forty hybrid models were developed by combining eight ML algorithms with four filter‐based feature‐selection methods and principal component analysis (PCA) as a dimensionality reduction method, and these models were evaluated using seven metrics. The weight coefficients for these metrics are determined using the 2‐tuple Linguistic Fermatean Fuzzy‐Weighted Zero‐Inconsistency (2TLF‐FWZIC) method. The Vlsekriterijumska Optimizcija I Kompromisno Resenje (VIKOR) approach is applied to rank the developed models. According to 2TLF‐FWZIC, classification accuracy (CA) and precision obtained the highest importance weights of 0.2439 and 0.1805, respectively, while F1, training time, and test time obtained the lowest weights of 0.1055, 0.0886, and 0.1111, respectively. The benchmarking results revealed the following top‐performing models: the Gini index with logistic regression (GI‐LR), the Gini index with a decision tree (GI_DT), and the information gain with a decision tree (IG_DT), with VIKOR Q score values of 0.016435, 0.023804, and 0.042077, respectively. The proposed MCDM framework is assessed and examined using systematic ranking, sensitivity analysis, validation of the best‐selected model using two unseen trauma datasets, and mode explainability using the SHapley Additive exPlanations (SHAP) method. We benchmarked the proposed methodology against three other benchmark studies and achieved a score of 100% across six key areas. The proposed methodology provides several insights into the empirical synthesis of this study. It contributes to advancing medical informatics by enhancing the understanding and selection of trauma injury detection models for ICUs.

Anomaly detection in videos is challenging due to the complexity, noise, and diverse nature of activities such as violence, shoplifting, and vandalism. While deep learning (DL) has shown excellent performance in this area, existing approaches have struggled to apply DL models across different anomaly tasks without extensive retraining. This repeated retraining is time‐consuming, computationally intensive, and unfair. To address this limitation, a new DL framework is introduced in this study, consisting of three key components: transfer learning to enhance feature generalization, model fusion to improve feature representation, and multitask classification to generalize the classifier across multiple tasks without training from scratch when a new task is introduced. The framework’s main advantage is its ability to generalize without requiring retraining from scratch for each new task. Empirical evaluations demonstrate the framework’s effectiveness, achieving an accuracy of 97.99% on the RLVS (violence detection), 83.59% on the UCF dataset (shoplifting detection), and 88.37% across both datasets using a single classifier without retraining. Additionally, when tested on an unseen dataset, the framework achieved an accuracy of 87.25% and 79.39% on violence and shoplifting datasets, respectively. The study also utilises two explainability tools to identify potential biases, ensuring robustness and fairness. This research represents the first successful resolution of the generalization issue in anomaly detection, marking a significant advancement in the field.

Fracture detection in medical imaging is crucial for accurate diagnosis and treatment planning in orthopaedic care. Traditional deep learning (DL) models often struggle with small, complex, and varying fracture datasets, leading to unreliable results. We propose FracNet, an end-to-end DL framework specifically designed for bone fracture detection using self-supervised pretraining, feature fusion, attention mechanisms, feature selection, and advanced visualisation tools. FracNet achieves a detection accuracy of 100% on three datasets, consistently outperforming existing methods in terms of accuracy and reliability. Furthermore, FracNet improves decision transparency by providing clear explanations of its predictions, making it a valuable tool for clinicians. FracNet provides high adaptability to new datasets with minimal training requirements. Although its primary focus is fracture detection, FracNet is scalable to various other medical imaging applications. • Propose an end-to-end deep learning framework for accurate bone fracture detection. • Incorporate attention mechanisms to enhance deep network performance. • Use feature fusion and selection to improve representation and model robustness. • Evaluate on three datasets, outperforming state-of-the-art methods. • Validate single classifier generalisation across multiple datasets successfully.

Deep learning (DL) in orthopaedics has gained significant attention in recent years. Previous studies have shown that DL can be applied to a wide variety of orthopaedic tasks, including fracture detection, bone tumour diagnosis, implant recognition, and evaluation of osteoarthritis severity. The utilisation of DL is expected to increase, owing to its ability to present accurate diagnoses more efficiently than traditional methods in many scenarios. This reduces the time and cost of diagnosis for patients and orthopaedic surgeons. To our knowledge, no exclusive study has comprehensively reviewed all aspects of DL currently used in orthopaedic practice. This review addresses this knowledge gap using articles from Science Direct, Scopus, IEEE Xplore, and Web of Science between 2017 and 2023. The authors begin with the motivation for using DL in orthopaedics, including its ability to enhance diagnosis and treatment planning. The review then covers various applications of DL in orthopaedics, including fracture detection, detection of supraspinatus tears using MRI, osteoarthritis, prediction of types of arthroplasty implants, bone age assessment, and detection of joint-specific soft tissue disease. We also examine the challenges for implementing DL in orthopaedics, including the scarcity of data to train DL and the lack of interpretability, as well as possible solutions to these common pitfalls. Our work highlights the requirements to achieve trustworthiness in the outcomes generated by DL, including the need for accuracy, explainability, and fairness in the DL models. We pay particular attention to fusion techniques as one of the ways to increase trustworthiness, which have also been used to address the common multimodality in orthopaedics. Finally, we have reviewed the approval requirements set forth by the US Food and Drug Administration to enable the use of DL applications. As such, we aim to have this review function as a guide for researchers to develop a reliable DL application for orthopaedic tasks from scratch for use in the market.

The degree of compaction in the levee building materials is a crucial factor that affects the piping phenomena. The density and compaction of the soil strata determine the structural soundness of the levee. Segments with reduced density or compaction can become weak spots during floods. To assess part of the Helena levee (2,500 m) in Arkansas (AR), the United States, an extensive ground-penetrating radar (GPR) fieldwork was conducted. This reliable method will undoubtedly improve the assessment procedure of the levee structure by identifying the weak spots within the structure that result from poor compaction of the levee core layers in a short time with high accuracy. However, interpreting the GPR data can be challenging and requires specialised knowledge. Obtaining meaningful insights typically involves a time-consuming process of extensive manual processing and visual inspection. To address this issue, this article proposes a novel, reliable deep-feature fusion framework for GPR data to identify horizontal variation in the soil compaction of a levee. To address data scarcity, a new type of transfer learning in the same domain is adopted, and four deep learning models (Xception, Inception, EfficientNet and MobileNet) are used to extract features. The combined features are then used to train and test five machine learning classifiers (Neural Network, Support Vector Machine, K-Nearest Neighbour, Logistic Regression, and Naive Bayes). The best combination of deep Learning and machine learning is four models with the neural network classifier which achieved the highest results by obtaining an accuracy of 98.2%, an F1 score of 97.6%, and an area under the curve of 99.9%. The proposed framework faced an additional challenge when subjected to an unseen dataset of 1,511 images reserved primarily for testing. Remarkably, it achieved an accuracy rate of 95.7% with the neural network classifier. This article presents a new research direction that has substantial potential in various domains, including civil engineering, the petroleum sector, road safety, agriculture, and more.

Early diagnosis of lung cancer is critical as it can save people’s lives. Long-range dependencies within volumetric medical images are essential attributes for accurate lung nodule classification. Many deep learning-based methods are used for lung nodule classification; however, the construction of the lung nodule is not axial and can be any shape. Thus, the nodules are interrelated through their adjacent slices axially and diagonally, locally and globally, making their capture through local convolutional operations challenging. In this article, we benefited from local co-occurrences of texture features and graph neural networks (GNNs) to effectively capture important patterns considering the long-range dependencies among adjusted slices. The proposed framework comprises a multi-side graph construction layer (MSGCL) that computes informative texture features and captures spatial relationships from cross-sectional and longitude orientations of the nodule, creating two sets of nodes. Further, the graph-based fusion of long-range dependency layers (GFLL) is used to deeply fuse and generate attentive edges among fused nodes. The LIDC-IDRI dataset, used for training and testing the proposed Multi-side Graph Neural Network-based Attention for Local Co-occurrence Features Fusion (MS-GNN-ALCFF), achieves state-of-the-art. The LUNGx dataset used as an unseen dataset shows that our model is generalisable compared to baselines. The results of the proposed method when the LIDC-IDRI dataset was used for train and testing were 87.17 ± 0.84 %, 91.01 ± 1.14 %, 88.6 ± 1.42 %, 89.7 ± 0.87 %,95 ± 1.22 % and 76.7 ± 1.4 % in terms of Accuracy, Precision, Recall, F1-score, AUC and MCC respectively. When the LUNGx dataset is used for testing as an unseen dataset, the results were 69.86 ± 2.4 %, 75 ± 2.7 %, 71.4 ± 3.1 %, 73.17 ± 3.7 %, 70.2 ± 1.36 % and 71 ± 3.8 % for Accuracy, Precision, Recall, F1-score, AUC and MCC respectively. These findings represent the significance of the ability of GNNs to construct a multi-set of nodes through the proposed MSGCL layer and fuse deeply through the proposed GFLL layer.

Deep learning has emerged as a powerful tool in various domains, revolutionising machine learning research. However, one persistent challenge is the scarcity of labelled training data, which hampers the performance and generalisation of deep learning models. To address this limitation, researchers have developed innovative methods to overcome data scarcity and enhance deep model learning capabilities. Two prevalent techniques that have gained significant attention are transfer learning and self-supervised learning. Transfer learning leverages knowledge learned from pre-training on a large-scale dataset, such as ImageNet, and applies it to a target task with limited labelled data. This approach allows models to benefit from the learned representations and effectively transfer knowledge to new tasks, resulting in improved learning performance and generalisation. On the other hand, self-supervised learning focuses on training models using pretext tasks that do not require manual annotation, allowing them to learn valuable representations from large amounts of unlabelled data. These learned representations can then be fine-tuned for downstream tasks, mitigating the need for extensive labelled data. In recent years, transfer and self-supervised learning have found applications in various fields, including medical image processing, video recognition, and natural language processing. These approaches have demonstrated remarkable achievements, enabling breakthroughs in areas such as disease diagnosis, object recognition, and language understanding. However, while these methods offer numerous advantages, they also have limitations. For example, transfer learning may face domain mismatch issues between the pre-training and target domains, while self-supervised learning requires careful design of pretext tasks to ensure meaningful representations. This review paper explores the recent applications of these pre-training methods in various fields within the past three years. It delves into the advantages and limitations of each approach, assesses the performance of models employing these techniques, and identifies potential directions for future research. By providing a comprehensive review of current pre-training methods, this article offers guidance for selecting the best technique for specific deep learning applications to address the data scarcity issue.

Abstract Meeting the rising global demand for healthcare diagnostic tools is crucial, especially with a shortage of medical professionals. This issue has increased interest in utilizing deep learning (DL) and telemedicine technologies. DL, a branch of artificial intelligence, has progressed due to advancements in digital technology and data availability and has proven to be effective in solving previously challenging learning problems. Convolutional neural networks (CNNs) show potential in image detection and recognition, particularly in healthcare applications. However, due to their resource-intensiveness, they surpass the capabilities of general-purpose CPUs. Therefore, hardware accelerators such as application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), and graphics processing units (GPUs) have been developed. With their parallelism efficiency and energy-saving capabilities, FPGAs have gained popularity for DL networks. This research aims to automate the classification of normal and abnormal (specifically Diabetic Foot Ulcer—DFU) classes using various parallel hardware accelerators. The study introduces two CNN models, namely DFU_FNet and DFU_TFNet. DFU_FNet is a simple model that extracts features used to train classifiers like SVM and KNN. On the other hand, DFU_TFNet is a deeper model that employs transfer learning to test hardware efficiency on both shallow and deep models. DFU_TFNet has outperformed AlexNet, VGG16, and GoogleNet benchmarks with an accuracy 99.81%, precision 99.38% and F1-Score 99.25%. In addition, the study evaluated two high-performance computing platforms, GPUs and FPGAs, for real-time system requirements. The comparison of processing time and power consumption revealed that while GPUs outpace FPGAs in processing speed, FPGAs exhibit significantly lower power consumption than GPUs.

Given the tremendous potential and influence of artificial intelligence (AI) and algorithmic decision‐making (DM), these systems have found wide‐ranging applications across diverse fields, including education, business, healthcare industries, government, and justice sectors. While AI and DM offer significant benefits, they also carry the risk of unfavourable outcomes for users and society. As a result, ensuring the safety, reliability, and trustworthiness of these systems becomes crucial. This article aims to provide a comprehensive review of the synergy between AI and DM, focussing on the importance of trustworthiness. The review addresses the following four key questions, guiding readers towards a deeper understanding of this topic: (i) why do we need trustworthy AI? (ii) what are the requirements for trustworthy AI? In line with this second question, the key requirements that establish the trustworthiness of these systems have been explained, including explainability, accountability, robustness, fairness, acceptance of AI, privacy, accuracy, reproducibility, and human agency, and oversight. (iii) how can we have trustworthy data? and (iv) what are the priorities in terms of trustworthy requirements for challenging applications? Regarding this last question, six different applications have been discussed, including trustworthy AI in education, environmental science, 5G‐based IoT networks, robotics for architecture, engineering and construction, financial technology, and healthcare. The review emphasises the need to address trustworthiness in AI systems before their deployment in order to achieve the AI goal for good. An example is provided that demonstrates how trustworthy AI can be employed to eliminate bias in human resources management systems. The insights and recommendations presented in this paper will serve as a valuable guide for AI researchers seeking to achieve trustworthiness in their applications.