K. B. V. Brahma Rao

Cardiovascular disease (CVD) impairs the functioning of the heart and blood vessels, often leading to severe physical disability. Early and automated detection of CVD can therefore play an important role in saving lives. Although many studies have addressed this issue, there remains scope for improving accuracy and reliability. In this research, a deep learning (DL)-based framework is developed for CVD prediction. Initially, the retinal images are preprocessed, and the optic disc is identified using a pyramid scene passing network (PSP-Net) optimized by the fractional jellyfish search optimization (FJSO). Then, blood vessel segmentation is performed using an FJSO-optimized spatial attention U-Net (SA-UNet). After segmentation, the statistical features are extracted, and textural features are derived from the input data. Both sets of features contribute to the CVD detection process. Furthermore, CVD risk prediction is carried out using an FJSO-enhanced deep Q-Net (FJSO-Deep Q-Net). Experimental evaluation shows that the proposed method achieves superior classification performance, with accuracy, sensitivity, and specificity values of 0.934, 0.924, and 0.939, respectively.

To address reliance on manual triage in current teleconsultation platforms, we propose a BTB intelligent triage model that supports next-generation 6G healthcare systems by integrating multi-feature fusion and Explainable AI (XAI) techniques. Assuming that the assigned consultation department in the dataset represents the optimal choice, the model uses patient age, gender, and preliminary diagnoses from primary care physicians to recommend appropriate specialty departments. To capture both local textual features and contextual dependencies in unstructured teleconsultation application records, the model combines BERT-based word embeddings, Text Convolutional Neural Network (TextCNN), and a double-layer Bidirectional Gated Recurrent Unit (BiGRU) network. The extracted features are fused and passed through a fully connected layer to output a ranked list of recommended departments. The model was trained and validated on large-scale data from the National Telemedicine Center. When recommending the top five departments, it achieved an accuracy of 90.77% and a Mean Reciprocal Rank (MRR) of 76.45%. Moreover, the incorporation of explainability enables clinicians and patients to understand the rationale behind AI-driven recommendations, fostering trust and transparency. Experimental comparisons confirm that the proposed model significantly outperforms traditional machine learning approaches, offering a robust and interpretable solution for intelligent triage in teleconsultation-based consumer healthcare platforms.

With the rapid development of digital twin technology and the sixth generation (6G), holographic communication is emerging as a transformative feature in Digital twin, offering immersive experiences in smart education, entertainment, healthcare, and beyond. However, existing network resource scheduling methods often fall short in addressing the massive bandwidth, ultra-low latency, and strict data synchronization requirements essential for high-fidelity holographic services in 6G-enabled environments. To address these challenges, this paper explores a novel intelligent computing-integrated network (ICIN) architecture designed specifically for holographic communication in 6G-based Digital twin. Leveraging the inherent in-network computing capabilities of ICINs, the proposed system performs real-time computation and optimization during data transmission, effectively reducing latency and alleviating bandwidth constraints. Building upon prior research in intelligent identifier-based networking, we introduce a dynamic identifier mapping mechanism based on service identifiers, resource chain identifiers, network function identifiers, and component identifiers. This enables adaptive scheduling of heterogeneous network resources, such as computing, storage, and forwarding, tailored to the diverse demands of 6G-driven holographic services. A prototype system was developed using this approach, and experimental results demonstrate that, compared with traditional edge computing methods, the proposed ICIN-based system reduces network bandwidth load by 60% and transmission delay by 45%, while providing enhanced flexibility for resource orchestration in next-generation consumer 6G communication scenarios.

The increasing use of federated learning (FL) in healthcare IoT demands rigorous verification to ensure the correctness of remote model training without compromising patient data privacy. However, existing approaches either assume full trust in clients or introduce high computational and communication costs when integrating cryptographic guarantees. In this work, we propose a lightweight, privacy-preserving federated learning framework that integrates zk-SNARK-based verifiable training over a ring topology. Our system ensures that each client’s model update and aggregation step can be independently verified without revealing sensitive data or requiring a central auditor. We design an efficient proof composition strategy (CGro16) tailored for chained convolution operations and commitment schemes optimized for healthcare models. We also introduce a matrix polynomial-based masking mechanism (MatProofs) to support zero-knowledge commitments for convolutional neural networks (CNNs).Experimental results on standard benchmarks (MNIST, CIFAR-100) show up to 47% reduction in proof generation time and 39% lower memory overhead compared to baseline zk-SNARK schemes. The protocol is also benchmarked on edge devices (Jetson Nano, Raspberry Pi), confirming its suitability for remote and wearable healthcare scenarios.

Digital forensics and biometric authentication play a crucial role in ensuring security in consumer electronics, particularly in intelligent wearable devices. These devices collect users’ health data and transmit it to cloud servers via wireless channels, exposing them to computational delays and security threats. While fog computing enhances cloud computing by improving response efficiency, it also introduces higher security requirements. This paper proposes a secure and efficient Authentication and Key Agreement (AKA) protocol tailored for biometric systems in consumer electronics, leveraging the Elliptic Curve Diffie-Hellman (ECDH) problem. The protocol integrates smart cards, passwords, and biometric authentication techniques, employing one-way hash functions and elliptic curves to safeguard user anonymity. Additionally, it utilizes random numbers and timestamps to ensure the integrity of transmitted messages, enabling mutual identity authentication and establishing a three-party session key. An informal security analysis confirms that the protocol can withstand various known cyber threats, while its robustness is further validated using the Random Oracle (ROR) model and the ProVerif formal security tool. Comparative analysis with existing protocols reveals that the proposed method reduces computational overhead to 66% of conventional solutions while offering superior security and performance, making it highly suitable for secure biometric systems in consumer electronics and digital forensic applications.

With the rapid evolution of holographic technology, its integration into IoT-based consumer electronics is revolutionizing user experiences in areas such as smart entertainment systems, immersive education platforms, and remote healthcare solutions. However, existing network resource scheduling approaches fall short of meeting the stringent requirements of holographic communication, which demands ultra-high bandwidth, low latency, and synchronized data transmission. To address these challenges, this paper introduces an intelligent computing-integrated network architecture tailored for consumer electronics, leveraging in-network computing to dynamically process and optimize data during transmission. We propose a novel identification mapping framework based on service, resource chain, network function, and network component identifiers to enable adaptive and efficient scheduling of holographic services across heterogeneous resources, including compute, storage, and forwarding elements. Building on this, an intelligent computing-integrated network system is developed to support seamless holographic communication in IoT consumer electronics. Experimental evaluations reveal that the proposed system reduces network bandwidth consumption by up to 60% and transmission latency by 45%, outperforming conventional edge computing approaches and significantly enhancing the responsiveness and efficiency of consumer electronic devices.

In order to address the issue of insufficient task offloading decisions in vehicle networks of transportation cyber-physical systems (TCPS) because of multitasking and resource constraints, this study presents a quasi-Newton deep reinforcement learning-based two-stage online offloading (QNRLO) algorithm. Computer simulation experiments show that the approach performs exceptionally well in terms of convergence under various conditions and parameter configurations. Most of the trials are carried out in a simulated setting, and further real-world scenarios may be required to confirm the algorithm’s efficacy. This methodology initially implements batch normalization techniques to enhance the training process of the deep neural network, subsequently utilizing the quasi-Newton method for optimization to successfully approximate the ideal answer. According to the experimental results, the QNRLO algorithm’s loss function and normalized computation rate have converged after 2,000 iterations, demonstrating the algorithm’s excellent stability and dependability. The findings demonstrate that the computational load and training time can be further optimized by appropriately adjusting certain parameters without compromising convergence performance. Furthermore, the technique incorporates system transmission time allocation into the TCPS model, hence augmenting the model’s practicality. The proposed approach markedly enhances the efficiency and stability of job offloading compared to previous algorithms, effectively addressing task offloading challenges in TCPS and exhibiting considerable applicability and reliability.

Protein sequence classification needs to be performed quickly and accurately to progress bioinformatics advancements and the production of pharmaceutical products. Extensive comparisons between large databases of known proteins and unknown sequences are necessary in traditional protein classification methods, which can be time‐consuming. This labour‐intensive and slow manual matching and classification method depends on functional and biological commonalities. Protein classification is one of the many fields in which deep learning has recently revolutionized. The data on proteins are organized hierarchically and sequentially, and the most advanced algorithms, such as Deep Family‐based Method (DeepFam) and Protein Convolutional Neural Network (ProtCNN), have shown promising results in classifying proteins into relative groups. On the other hand, these methods frequently refuse to acknowledge this fact. We propose a novel hybrid model called ProteinCNN‐BLSTM to overcome these particular challenges. To produce more accurate protein sequence classification, it combines the techniques of amino acid embedding with bidirectional long short‐term memory (BLSTM) and convolutional neural networks (CNNs). The CNN component is the most effective at capturing local features, while the BLSTM component is the most capable of modeling long‐term dependencies across protein sequences. Through the process of amino acid embedding, sequences of proteins are transformed into numeric vectors, which significantly improves the precision of prediction and the representation of features. Using the standard protein samples PDB‐14189 and PDB‐2272, we analyzed the proposed ProteinCNN‐BLSTM model and the existing deep‐learning models. Compared to the existing models, such as CNN, LSTM, GCNs, CNN‐LSTM, RNNs, GCN‐RNN, DeepFam, and ProtCNN, the proposed model performed more accurately and better than the existing models.