Skin diseases constitute one of the most prevalent dermatological disorders worldwide and their early and precise diagnosis is essential for reducing diagnostic delays as well minimizing disease associated mortality. Motivated by this clinical need this work proposed a novel hybrid skin lesion classification framework termed ResBR-XGBR is proposed, which integrates deep feature extraction using ResNet50, hybrid feature optimization and ensemble decision fusion using Random Forest and XGBoost classifiers for the multiclass skin lesion classification using HAM10000 dataset. To capture rich hierarchical features, two pretrained deep convolutional architectures, ResNet50 and EfficientNet-B0 were employed. Subsequently, a multi-stage feature optimization strategy incorporating: Boruta, Chi-square, and Recursive Feature Elimination (RFE) was employed to remove redundant and irrelevant feature while retaining only the most discriminative, statistically, and diagnostically features. The selected features were further evaluated on multiple classifiers using an 80:20 training testing split with Grid Search Cross validation to perform hyperparameter tuning and identify the optimal learning for each model. In scenario 1, the feature subsets obtained through Boruta and Recursive Feature Elimination (RFE) demonstrated superior compatibility with Random Forest, XGBoost, and their ensemble classifier, where the best accuracy of 91.4%±1.2% was achieved using ResNet50 features. Motivated by these findings, scenario 2 involved the fusion of Boruta and RFE selected features, followed by re-evaluation across individual and ensemble classifiers. The experimental results revealed that the proposed hybrid ensemble framework (ResBR-XGBR) consistently outperformed all standalone learners and attained the highest test classification accuracy of 95.2%±0.5%.

Relevant data is helpful in aiding a clinician in decision making, which requires the capture of information efficiently. However, retrieval of responses from various sources like electronic health records (EHRs) and other related documents is still a difficult task. In solving this problem, a system called Natural Language Querying for Clinical Information Extraction (NLQ-CIE) has been developed using a trained model of PubMed BERT which answers user queries in a suitable and context-dependent manner. In this model, a clinician is responsible for providing a patient’s medical data alongside pertinent queries from which answers need to be fetched. The system manages the questions using tokenization and word embeddings, and the medical information undergoes a set of procedures including data cleansing, tokenization, and standardization. Afterwards, these models are trained for the first time using the Biomedical Clinical Question Answering dataset (BCQA) which has been aimed towards supporting the biomedical question answering process. This guarantees that relevancy to the biomedical NLP domain is preserved, as well as providing the opportunity to assess the model performance. Thereafter, model specialization on the defined tasks is completed to enhance the model's capability of retrieving patient clinical data efficiently. During the extraction phase, quality assessment is performed relating to the data validation. For measuring model performance, some relevant model evaluation metrics like Exact Match (EM) and F1 Score are applied, proving the usefulness of the system. Using domain knowledge for clinical use and customizing PubMed BERT, the system proves useful in extracting important clinical information and thereby aiding clinicians to make better decisions.

In the rapidly evolving field of biomedical research, the ability to efficiently and accurately summarize vast amounts of text is essential. Hallucination in the context of summarization refers to the generation of information that is not present in the source text. Generating an abstractive summary poses a significant challenge, especially in domains like biomedicine, where it’s difficult to balance the accuracy with other summary characteristics, such as readability and contextual understanding. This paper proposes a HE2A_BTS model for generating abstractive summaries of biomedical research articles that balances factual accuracy with linguistic clarity. The model utilizes BioBERT for extractive summarization in the first phase, followed by a stacked LSTM with a UMLS (Unified Medical Language System) based Attention layer and copying mechanism for abstractive summarization in the second phase. The performance evaluation of the proposed hybrid abstractive summarization model is based on multidimensional analysis, which combines both automated evaluation metrics and human-expert scoring. These include BERTScore (78.07) for semantic alignment, ROUGE (60.10) for n-gram and structural overlap, and simulated FactCC (83.5) for factual consistency. The comprehensive evaluation indicates that HE2A_BTS is effective in generating semantically faithful, readable, and factually consistent biomedical summaries that retains the complex biomedical terms. By balancing linguistic simplification with information integrity validated through multiple metric families, the model shows promise for biomedical text summarization.

There are issues of interoperability, privacy and security in the current Electronic Health records (EHR) system. To overcome these challenges the authors have proposed a patient-centric multichain Decentralized Health Data Management (DHDM) system. This system makes use of the multichain framework of the polygon to provide interoperability across the authorized hospitals and polygon with a multichain structure. It acts as a conduit for external systems information, while also enabling the deployment of health records on the Ethereum blockchain with smart contracts. With focus on fully decentralization of the current medical healthcare system, DHDM aims to address the scalability and cost issues that are sometimes connected with the blockchain based apps by storing records on the Interplanetary file system (IPFS). This work also emphasizes the current challenges of the EHR system, which includes security, privacy, and interoperability difficulties despite its widespread use. When data is shared among the various healthcare institutes, it not only affects the price but also the quality of the data. The authors have proposed a secure means of collecting, obtaining and uploading data by putting a decentralized DHDM on blockchain. This method would improve the way medical data is being managed, which would increase the healthcare system overall efficiency.

The use of image-sensing and real-time processing in Intelligent Transportation Systems (ITS) has introduced a sudden surge in transmitting and gathering high-resolution visual data from vehicle cameras, road infrastructures, and user devices. Such image data are, however, exceedingly susceptible to interception, tampering, and privacy breaches with regard to imminent quantum computing attacks that can break classical encryption algorithms. With such constraints in view, the paper presents a new Hybrid Quantum-Classical Image Encryption Framework that integrates chaos-based bit-level image encryption and quantum-resistant encryption measures to ensure high-security protection of image information in ITS infrastructures. The new framework integrates a customized bit-level chaotic permutation scheme using a Rearranged Arnold Cat Map (R-ACM) and 2D Logistic-Sine Chaotic Maps for confusion and diffusion, and the inclusion of a Quantum Key Distribution (QKD) or post-quantum lattice-based Kyber Key Encapsulation Mechanism (KEM) for secure key negotiation. The two-pyramidal security architecture enhances sensitivity to key and plaintext variations, offers chosen-plaintext, differential, noise, and occlusion attack immunity, and supports efficient encryption of RGB and grayscale image information without excessively large time overhead. Experimental results on representative ITS-relevant image data sets verify superior performance with mean NPCR > 99.60%, UACI ≈ 33.5%, entropy measures close to 8.0, and significantly suppressed correlation between neighboring pixels. Further, key space analysis demonstrates a combinatorial complexity of over 2²⁵⁶, making brute-force and quantum-type attacks computationally infeasible. The new framework is extremely suitable for real-time implementation in autonomous vehicles, roadside edge nodes, and intelligent traffic monitoring systems, thereby enabling secure, intelligent, and privacy-preserving ITS infrastructure in the post-quantum era.

With an increasing emphasis on user data control and privacy regulations, such as the General Data Protection Regulation (GDPR), machine unlearning (MU) has emerged as a crucial mechanism for managing data in AI systems. MU enables models to remove the influence of specific user data upon request. This problem becomes more complex in collaborative settings such as Federated Learning (FL), where data remains distributed across multiple clients, giving rise to Federated Unlearning (FU). In large-scale deployments, particularly those supported by cloud infrastructure, retraining models to satisfy data deletion requests can be computationally expensive, energy-intensive, and disruptive to ongoing services. This highlights the importance of having effective methods to unlearn outdated practices that can hinder the growth of a system and compromise data privacy awareness. We propose PRUNE-FL (Privacy-preserving Retention-focused Unlearning with Neuro-Evolution in Federated Learning), a framework that uses relevance-guided pruning and evolutionary optimization to delete the influence of the targeted data. PRUNE-FL is different from methods that rely on retraining or coarse parameter updates because it focuses on finding and changing the parameters that are most closely related to the data that needs to be forgotten. The approach is based on a synaptic relevance scoring system to figure out how each model parameter relates to certain client or class-level data. This makes it easier to find parameters that are connected to the target data. Then, the unlearning task is set up as a multi-objective problem to find a balance between the overall performance and the forgetting of unnecessary information. Finally, a genetic algorithm is used to implement an evolutionary pruning strategy. It seeks the optimal pruning settings that operate within the constraints of federated learning. Thus, PRUNE-FL helps in unlearning specific data without having to retrain the whole system. Tests on the CIFAR-10 dataset in both IID (Independent and Identically Distributed) and non-IID settings show that PRUNE-FL has higher accuracy. It also effectively removes the influence of the targeted data. The results also show it to be strong against bad patterns like backdoor triggers. Overall, PRUNE-FL enhances privacy by selectively unlearning the data and using fewer resources in federated environments.

The existing healthcare claims process suffers from many ills, like fraud, huge paperwork, and delays in claims verification, affecting both patients and insurers badly. This paper represents a blockchain-based health insurance system based on Ethereum smart contracts and IPFS to enhance the security and transparency and ensure automation in claims processing. The proposed system consists of the blockchain layer for automating the policy management, claims processing, and fraud detection via Policy, Claim, and Audit smart contracts; a decentralized storage layer (IPFS) that ensures secure and tamper-proof storage of claim-related documents; and a payment layer that formally settles transactions via smart contracts, thus lessening the company's reliance on intermediaries. Integrating middleware APIs ensures good interoperability among insurers, hospitals, patients, and administrators-real-time tracking of claims and compliance with regulatory demands. This will be a global, scalable, fraud-resistant model that optimizes claims settlement and minimizes administrative overhead and maximizes user satisfaction. This paper underscores how health insurance can benefit from blockchain technology.

For the speech and hearing-impaired population, sign language is an essential means of communication. The accessibility of sign language, however, makes it extremely challenging to employ effectively with those who don't comprehend it. This paper proposes to create a comprehensive system to translate sign language into both text and speech to overcome this communication gap. The system makes use of the convolutional neural network (CNN) technique to identify hand gestures made by people from images. The main purpose is to recognize hand motions from a camera image. The position of the hand and its orientation are then applied to obtain the training and testing data for the CNN. Firstly, the hand is detected from the image captured by the webcam then the image-processing Mediapipe library is used. After passing through a filter, this is sent to a classifier, which makes predictions about the class of hand movements. The CNN model is then trained using these pictures. This strategy could be advantageous to the hearing communities as well as the deaf/dumb people. Deaf/dumb individuals can express themselves more effectively than hearing individuals, while the latter can understand and respond to sign language gestures through text and speech. This paper explores the customization of sign language recognition for regional and cultural variations, making it adaptable to diverse sign languages worldwide.

The increasing demand for healthier food alternatives has highlighted the need for nutritional intelligence in evaluating food quality. For this purpose, the nutritional and biochemical characteristics of each food ingredient should be classified to develop intelligent food recommendation systems and to help individuals make informed dietary choices. This study uses nutritional and biochemical data to investigate the applicability of machine learning techniques in the categorization of food ingredients into healthy and unhealthy. Six algorithms namely, Decision Tree, Random Forest, Support Vector Machine, Logistic Regression, K-Nearest Neighbors, and Extreme Gradient Boosting were implemented and compared to assess their predictive capabilities. The experimental results indicate that XGBoost outperformed all other models, with an average accuracy of 94%, demonstrating its better generalization and predictive power based on the ensemble gradient boosting technique. Random Forest and KNN have also achieved a remarkable 92% accuracy owing to ensemble learning and distance-based decision limits, respectively. The findings demonstrate how machine learning can be effectively used to advance nutrition intelligence, as well as providing a comprehensive and extensible foundation for incorporating computational approaches into current food assessments and tailored healthcare.

Spices should be authentic for customers to justify consumption, but counterfeiting, and increasing prices are challenges in spices sector. This is compounded by the absence of information on the full manufacturing of spices and the labelling problem. In response to this issue, the present study intends to develop a decentralized blockchain system. It will expand customer convenience by allowing them to view and monitor the flow of spices/herbs from the farmer to the bowl. The information that it will include are of origin, quality control measures, storage and roasting sites, as well as manufacturing and expiration dates. Further, the system will be able to provide transparency at all the levels and provide feedback generated using machine learning algorithms. Through it, the companies would be free from fraudulent products, customers, etc.; they would also give customers a guarantee that the spices/herbs they are buying are genuine products.

Organ donation is a noble act providing possibility of new life and improves health plus well-being of society. There is an increasing number of patients who dies due to lack of transplantable organs. Blockchain technology can be a key solution for the needs by enabling transparency, immutability, dependability, and auditable computing within the system via utilizing decentralized network of peers and ledger having provable time-stamped proof of owning and generation. Implementing blockchain technology in the organ donation system can improve data analytics, transparency, and security. Consequently, the blockchain-enabled organ donation system-the DONOCHAIN will become a platform for the donor as well as the receiver to securely manage the information they are sharing with transparency, immutability, distributed nature, and reliability. Eliminating inefficiency will result in a decrease in cost by performing transactions through smart contracts, between parties without centralized control. Using smart contracts will help in authentication of the users to make the system more trustworthy for both the donor and receiver. Modern analytics maximizes allocation and use, enabling optimal organ matching. While regulatory requirements are enforced by automated compliance systems, real-time monitoring ensures timely response. Smart transplants change organ donation by providing improved patient outcomes, more efficacy, and global accessibility.

The modern cyber-physical systems (CPS), which power smart manufacturing and autonomous vehicles, need sensor data for their real-time decision processes. The systems remain exposed to false data injection (FDI) attacks and insider calibration drift issues, and single points of failure in metrology validation. The present methods for sensor precision verification do not solve these problems which create security threats and operational dangers. A Proof-of-Metrology (PoM) system needs to be developed to solve this problem.The system uses a metrology-aware consensus framework which runs Raft Byzantine Fault Tolerant (BFT) protocol to verify nodes before transaction validation through metrological compliance verification. The PoM integrates three fundamental elements, which consist of smart contracts for standard calibration enforcement and a Byzantine Fault-Tolerant (BFT) consensus model using ISO 17025-compliant validators for robustness and ZK-SNARKs for private verification of sensor data. The evaluation of PoM took place in a simulated controlled environment through open-source tools with synthetic sensor data. Simulation results show PoM outperforming Hyperledger by detecting false data injection at a 98% rate while reducing consensus latency to 120 ms and minimizing the mean absolute error (MAE) in metrology validation to 0.01%. The results show that PoM enhances sensor data validation operations through faster processing times and better accuracy and dependability for CPS systems.

When a customer switches from one telecom service provider to another, it reduces the company's earnings, which makes customer churn a problem for most businesses. We use two main strategies to address this issue: first is determining the primary factors that are responsible for customer churn and the second is evaluating which individuals are most inclined to depart. For the purpose of learning insights from data and forecast the customer churn rate, the customer churn prediction model uses Exploratory Data Analysis (EDA) to analyse historical data from the telecom industry. Nevertheless, in this model, we will use the data of the customers who have already churned to generate the training dataset of the model. Formerly used Customer Churn Prediction models employed data based on the feedback provided by users or churned customers. Future forecasts will also make advantage of this data. It will offer better data from which to create a prediction model.

Polyp is an earlier stage of cancer development in gastro-intestinal tract. Despite the fact that numerous techniques for automatic segmentation and detection of polyps have been developed, it still remains an open problem. Examining color images captured by the board camera in endoscopy is contemplated the most reliable method for detectingdeveloping polyps. To automate this process, in this paper we present a modified U-Net Neural Network based algorithm. This algorithm is a binary classifier which labels the pixels as part of the polyp or not. This model can analyze the video of GItract frame by frame and produce the collection of frames with possibility of presence of polyp thus reducing the time required to process the GI-tract. Detecting, localizing, and segmenting polyps using Kvasir-SEG, an open-access dataset of colonoscopy images, we benchmark various modern day approaches in this work, assessing bothmethod speed and accuracy. This is attributed to outstanding performance of image classification compared to preceding techniques. The automatically identifying polyps serves to aid gastroenterologists during colonoscopies. While existing literature contains publications addressing the challenge of polyp detection, many of these systems remain confined toresearch settings and lack implementation for clinical use. Hence, we present the inaugural fully open-source automated polyp detection system, which not only outperforms the best- performing system documented in the literature but is also prepared for immediate clinical application.

Public procurement, a critical facet in acquiring goods and services, has garnered scholarly attention in e-procurement. However, its application in healthcare remains underexplored. This paper addresses this gap by proposing a decentralized blockchain-based e-procurement system in healthcare (B-ePS) to enhance transparency, reliability, and security. The study leverages the Ethereum blockchain and smart contracts to establish B-ePS, aiming to mitigate traditional e-procurement concerns. By emphasizing the blockchain's non-tampering characteristics, information traceability, and decentralization, the authors contribute a practical framework that goes beyond descriptive methods. The research reviews existing literature, identifies challenges, and recommends specific solutions within the B-ePS framework. This approach positions B-ePS at the forefront of advancing trustworthy e-procurement in healthcare, addressing security implications, lack of trust among stakeholders, and the absence of a fully reliable e-procurement solution. In summary, this paper focuses on the purpose of improving public procurement in healthcare through a decentralized, transparent, and secure e-procurement system. The methodology involves the implementation of B-ePS using Ethereum blockchain and smart contracts. The major findings highlight the system's potential to address longstanding issues in traditional e-procurement. The paper concludes with implications for advancing reliable e-procurement solutions in healthcare, emphasizing the transformative impact of blockchain technology.

Self-reports or professional interviews have typically been used to diagnose depression, although these methods often miss significant behavioral signals. Sometimes, people with depression may not express their feelings accurately, which can make it hard for psychologists to diagnose them correctly. We believe that paying attention to how people speak and behave can help us better identify depression. In real-life situations, psychologists can use different methods, like listening to how someone talks, their body language and change in their emotions while talking. To detect signs of depression more accurately authors presents MANOBAL, a system that analyzes voice, text, and facial expressions to detect depression. We use the DAIC-WoZ dataset, which was requested from the University of Southern California (UoS). We used this dataset for the multimodal depression detection model. Deep learning is challenged with such complicated data, therefore MANOBAL used a multimodal method. It uses elements from audio recordings, text, and facial expressions to predict both depression and its severity. This fusion has two advantages: first, it can substitute for uncertain data in one modality (such as voice) by using input from another (text, facial expressions). Second, it can give more weight to more dependable data sources, which improves accuracy. Small datasets are not very helpful when testing accuracy in fusion models, but MANOBAL overcomes this by exploiting DAIC-Woz dataset's transfer characteristics and increasing training labels. The initial results are encouraging, with a root mean square error of 0.168 for predicting depression severity. Experiments show the effectiveness of combining modalities. High-level features based on Mel Frequency Cepstral Coefficients (MFCC) give useful information on depression, but adding additional audio characteristics and facial action unit increases accuracy by 10

Whenever there is too much information out there, it is desirable to summarize. If humans are trying to create the summary, it will take lot of time. Now to make the problem of summarizing information easier and more effortless one can automate the summarization process which can reduce the time taken in creating summary. This is called as automatic summarization. The two ways of summarization are extractive summarization and abstractive summarization. Extractive summarization and its applications have been the subject of extensive research and have received state of art solution. But abstractive summarization still is a progressive field as it is difficult to create abstractive summary as humans do. Also, it is still a question i.e., how to evaluate the quality of a summary? therefore, this paper is a comprehensive survey on the dataset used with its details and statistics, analysis of various abstractive summarization techniques and important parameters for evaluating the quality of summary. Deep leaning based models have given new direction in this field. The author also focuses on problems and challenges faced in the generation of summary which are opening the future research scope in this domain.

Automatic summarization of text in the biomedical field includes the usage of natural language processing (NLP) and machine learning techniques to generate concise and coherent summaries of biomedical texts. The paper present a Biomedical text abstractive summarization tool that uses Stacked LSTM(Long Short Term Memory) including attention layer and copying mechanism. Also, the model uses UMLS(Unified Medical Language System) for better learning of medical terms so as to adapt the domain knowledge which plays very important role for preparing summary for domain specific text. Data is first pre-processed which involves removing irrelevant text and stop words and normalizing the text. Medical Vocabulary is added with the help of UMLS which is a large Biomedical Metathesarus. It is then passed to the seq2seq model i.e. the stacked LSTM.The encoder uses three layers of LSTM and single LSTM layer for decoder. The role is to convert the preprocessed text into word embedding which are then passed to the Attention layer and hence allows the Attention layer to identify the most crucial parts of the original text and the database documents. The pointer network uses the contextualized representation to generate a condensed representation of the database documents which further generates summary i.e. concise and informative. Earlier work is using only single evaluation metric while the highlight of our work is that the prepared summary is evaluated two metrics i.e. BLEU score and ROUGE score and has achieved 0.921 and 59.9

The latest data demonstrates that 34% of resumes, 5% of academic certificates, and 23% of job applications in South Asian countries contain anomalies. The non-verification rate of academic certificates is found to be around 12.9% in certain developed countries. This emphasizes the issues pertaining to fraud with respect to certificate verification. The above figures demonstrate the need for a secure system that maintains the trust and transparency of credentials. To resolve this, we offer a robust solution that is based on Blockchain Technology for certificate validation by combining features of AI and advanced machine learning algorithms to enhance the verification process and thereby reduce credential fraud. The AI system detects anomalies or tampering attempts within each block; it also analyzes the pattern in block data, then verifies consistency and identifies suspicious changes, which may indicate manipulation of data. The user database is maintained through MongoDB. When compared to other approaches, the emphasis is on robustness through established cryptographic methods, which maintains its own secure database of hashes. Such an integration of technologies results in this cohesive system, ensuring efficient and secure digital certificate management as a scalable and adaptable solution. The focus on fundamental elements of blockchain makes it distinct when compared to more complex or isolated methods in other existing research works.