Machine Learning Models for Blood Glucose Level Prediction in Patients with Diabetes Mellitus Reehana Shaik, Supriya Rayana, Vijay S. Karwande, Ajay Hanumanthu, Pradeep Arun Patil, Meghana Regulapati, Mohammad Noor Shaik, Arvind Bhaskar Sonawane, Devi Aruna Jyothi Bommareddy, Reshma Gayathri Kassetti, Karishma Mohammad, Chennu Reha Bhavani, Rudrani Banavatu Revolutionizing Drug Research and Personalized Medicine Through AI and Machine Learning, 2026 Diabetes mellitus remains a major global health challenge due to its rising prevalence, high morbidity, and the complexity of managing dynamic glycaemic fluctuations. Blood glucose prediction is central to preventing hypoglycaemia and hyperglycaemia and enabling proactive, individualised diabetes management. Recent advances in machine learning and deep learning allow high-dimensional data from continuous glucose monitoring systems, insulin delivery records, lifestyle information, and wearable devices to improve glucose forecasting beyond analytical methods. This chapter addresses data sources, modelling techniques, and the therapeutic significance of machine learning-based blood glucose prediction in diabetes management. It comprises short-term glucose forecasting, risk-based glycaemic event prediction, insulin titration decision support, and integration with mobile health platforms and clinical dashboards utilising supervised learning, deep learning, and hybrid time-series models.
Real-time crypt analysis and defense mechanisms in cloud computing services Jayita Moulick, Rajesh Phursule, Archana P. Kale, Varinder Singh Rana, Shubham Goswami, Vijay S. Karwande Journal of Discrete Mathematical Sciences and Cryptography, 2025 The expanding dependence on cloud computing administrations has required progressed components for real-time tomb examination and defense to defend delicate information. This paper proposes a novel Various leveled Unified Learning (HFL) system outlined to improve versatility and productivity in dispersed inconsistency discovery inside cloud situations. The HFL strategy structures the learning handle into numerous layers—local, territorial, and global—facilitating productive show conglomeration and diminishing communication overhead. By leveraging territorial conglomeration some time recently worldwide integration, the proposed HFL system optimizes asset utilization and moves forward versatility, pleasing a better number of clients with decreased idleness. In addition, the HFL approach coordinating real-time irregularity discovery and reaction instruments, guaranteeing that rising dangers are recognized and relieved expeditiously. Comparative examination with existing strategies such as Combined Averaging (FedAvg), Personalized Unified Learning (PerFedAvg), and Unified Exchange Learning (FedTL) illustrates that the HFL system offers predominant show exactness, diminished communication overhead, and speedier joining times. Also, the proposed strategy improves information security and exactness in identifying irregularities. This investigate underscores the potential of HFL as a vigorous arrangement for real-time tomb investigation and defense, clearing the way for more secure and adaptable cloud computing administrations.
Adaptive-Personalised Federated Deep Learning for Privacy-Aware NAFLD Screening ShivaKrishna Deepak Veeravalli, Pradeep A. Patil, Tina Porwal, Vijay S. Karwande, Anmol S. Budhewar, Bhushan Marutirao Nanche International Conference on Innovations in Intelligent Systems Advancements in Computing Communication and Cybersecurity Isac3 2025, 2025 Non-alcoholic fatty liver disease (NAFLD) moves nearly a quarter of global adult population, yet current diagnostic pathways still rely on resource-intensive ultrasonography or invasive biopsy. This study introduces an Adaptive-Personalised Federated Deep-Learning (A-P-FedDL) framework that enables collaborative, privacy-preserving prediction of NAFLD from routine clinical and laboratory variables collected at geographically dispersed hospitals. The method builds on a lightweight four-layer convolutional neural network trained under the FedAvg protocol and augments it with two novel components: client-similarity weighting, which dynamically scales each participant’s model update by the statistical distance between local and global feature distributions, and adaptive local-epoch scheduling that lengthens or shortens on-device training depending on convergence speed. Experiments were conducted on a cohort of 577 subjects (377 positive, 200 negative) from New Taipei City Municipal Hospital using stratified five-fold cross-validation repeated three times. A-P-FedDL achieved 94.2 % accuracy, 93.1 % sensitivity, 95.3 % specificity, an F1-score of 0.934, and an AUROC of 0.973, outperforming vanilla FedAvg-CNN by 2.7 percentage points and the best centralised baseline by 3.5 points. The framework also converged in 60 rounds and reduced perclient communication to 6.6 MB, representing a 41 % bandwidth saving. A Wilcoxon test confirmed statistical significance (p = 0.003). These findings demonstrate that personalised aggregation and adaptive training schedules can simultaneously enhance predictive performance and communication efficiency, paving the way for scalable deployment of edge-enabled liver-health screening in primarycare networks. Further, the modular design allows straightforward extension to additional biochemical markers and imaging features, supporting precision hepatology initiatives.
Towards Accurate Maritime Surveillance: A Hybrid CNN-Transformer Architecture for Ship Detection in SAR Imagery Seshendranath Balla Venkata, Anorgul Ashirova, Vijay S. Karwande, Aruna T M, Erkin Kholiyarov, Satyam Singh International Conference on Innovations in Intelligent Systems Advancements in Computing Communication and Cybersecurity Isac3 2025, 2025 Finding ships in water using high-resolution radar images is goal of synthetic aperture radar (SAR) ship detection, which works in any lighting or weather scenario. By taking advantage of ships' distinct scattering characteristics against water's surface, it is an essential tool for border security, maritime surveillance, and monitoring illicit fishing or trafficking. To accurately recognize ships in synthetic aperture radar (SAR) images in real-time, this study presents AMTNet, a new attention-enhanced multiscale transformer network. Despite abundance of structural information in SAR images, object detection remains a tough task due to speckle noise and diverse backdrops. suggested AMTNet integrates a ResNet-based multiscale convolutional backbone, a spatial attention module (SAM), a channel attention module (CAM), and transformer-based contextual modeling to solve these problems. In addition, a feature exchange system is implemented to address issue of domain gaps that might be generated by differences in environment and sensors in bi-temporal SAR images. To improve accuracy of change detection and localization, this Siamese design allows cooperative exploitation of spatial, temporal, and contextual signals. Compared to top models like MSDFF-Net, GLDet, and LEAD-YOLO, AMTNet outperforms them all on SSDD dataset, with metrics like detection speed (138.5 FPS), accuracy (94.60 percent), recall (95.15 percent), and F1-score (74.87%). Attention processes and feature exchange have been proven to play a key role in ablation research. Another proof of model's practicality is its resilience in face of occlusion and noise. With its excellent performance in a variety of SAR imaging settings and its high interpretability, AMTNet is a computationally economical and effective option for marine surveillance.
Metaheuristic-Tuned GraMNet Architecture for Enhanced Video-Based Anomaly Detection using UCF50 Dataset Seshendranath Balla Venkata, Connor H. Wong, Vijay S. Karwande, Divyaraj G N, Satyam Singh, Rajendra V. Patil International Conference on Innovations in Intelligent Systems Advancements in Computing Communication and Cybersecurity Isac3 2025, 2025 Abnormal activity detection in video surveillance has become a critical research domain, driven by the increasing demand for intelligent security systems in public and private environments. Abnormal activity recognition involves identifying unusual or deviant behaviors within a monitored environment, often using video surveillance, sensor data, or wearable devices. This process is crucial in requests such as security monitoring, healthcare, besides smart homes, where detecting anomalies like falls, unauthorized access, or erratic movements can prevent harm or trigger timely interventions. By leveraging deep learning models besides spatio-temporal analysis, systems can learn typical behavior patterns and flag activities that diverge from norm. This study presents a novel hybrid deep learning framework integrating GramNet (GraMNet) architecture with bio-inspired Greater Cane Rat Algorithm (GCRA) for hyperparameter tuning. UCF50 dataset, comprising 50 real-world action categories, was employed to train and validate proposed model. GraMNet architecture utilizes modular SubNets conFig.d in both serial and parallel forms, enabling efficient feature extraction while maintaining computational feasibility. GCRA, inspired by intelligent territorial and foraging behavior of Greater Cane Rats, was applied for hyperparameter optimization to improve convergence and model generalization. system incorporates transfer learning to mitigate challenges of data scarcity and boosts classification accuracy. Experimental evaluations demonstrate that proposed GraMNet-GCRA framework achieves superior performance across multiple metrics, with a peak accuracy of 96%, precision of 94%, recall of 93%, besides F1-score of 93.5%, outperforming state-of-the-art metaheuristics such as AOA, WOA, DMO, and ADMO. Moreover, it significantly reduces training time without compromising model efficacy. Visual analysis using ROC and loss curves confirms stable training dynamics and effective anomaly classification. This work highlights potential of combining structured CNN-based models with biologically-inspired optimization for enhanced abnormal activity recognition in video data.
Leveraging Speech Driven Patterns Multimodal Machine Learning Framework for Accurate Early Stage Parkinson's Disease Prediction - A Survey V.S. Karwande, Umesh B. Pawar, Omkar Pattnaik Proceedings 2024 2nd International Conference on Advanced Computing and Communication Technologies Icacctech 2024, 2024 PD is progressive neurodegenerative ailment that significantly affects life quality owing to death of dopamine-generating neurons in brain's area of substantia nigra. Symptoms include trouble writing, walking & conversing. Speech patterns, gait, and EEG (Electroencephalography) signals have recently been identified as biomarkers for early Parkinson's disease identification, with speech tests being particularly helpful because approximately 90 percent of Parkinson's sufferers exhibit speech issues. As the illness worsens, the patient's voice becomes much weaker necessitating the use of non-invasive voice analysis tools. This study looks uses of approaches to ML and DL to identify & predict PD based on a variety of symptoms, such as speech data, movement shaking, movement flexibility, and ease of everyday tasks. Predictive models will be trained and tested using SVM, Naive Bayes, K-NN, RF, Logistic Regression and Decision Trees are examples of machine learning classifiers. The goal to develop a ML model that accurate examine PD by analyzing speech recordings. This study uses advanced computer techniques to provide a viable tool for beginning identification and monitoring of PD using non-invasive speech analysis.
Patient Engagement and Satisfaction in Ai-Enhanced Healthcare Management Yogesh Rathore, Vandana Mishra Chaturvedi, Khadilkar Sujay madhukar, Vijay Suresh Karwande, Anil Haribhau Rokade, Yogesh Nagargoje International Conference on Artificial Intelligence for Innovations in Healthcare Industries Icaiihi 2023, 2023 This study looks into how patient involvement and satisfaction are affected by AI-enhanced healthcare administration. Using a descriptive methodology and secondary data gathered from reliable sources, the research takes an interpretive approach within a deductive paradigm. The findings of the investigation indicate that artificial intelligence (AI)-driven solutions, such as chatbots and virtual assistants, enhance patient engagement by providing easy access to healthcare knowledge and customized interactions. Additionally, patients are happy with the effectiveness and customized care that AI systems offer. Nonetheless, consumer happiness levels are influenced by the caliber of contacts and worries about the confidentiality of data. Transparency, dependability, data privacy, provider expertise, and education for patients all influence people's trust in AI technologies. For artificial intelligence integration to be successful, a technical assessment emphasizes the significance of algorithm efficiency, scalability, seamless integration, data protection, and continuous maintenance. Patient-centered design should be prioritized, stakeholders should be informed, and AI systems should be continuously monitored. Prospective investigations ought to concentrate on extended periods, tailored treatment schemes, psychological assistance, reducing bias, economies of scale, and innovative artificial intelligence approaches.
