Shakeel Ahmed
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Shakeel Ahmed has a total of 136 co-authors. Here are the top co-authors by number of shared publications:
- Abdulaziz Alhumam
Department of Computer Science, College of Computer Sciences and Information Technology, King Faisal University, Al-Ahsa, Saudi Arabia
11 shared publications - Parvathaneni Naga Srinivasu
Amrita School of Computing, Amrita Vishwa Vidyapeetham, India
9 shared publications Meenu Gupta
Chitkara University Institute of Engineering and Technology, Chitkara University
7 shared publicationsParvathaneni Naga Srinivasu
Amrita School of Computing, Amrita Vishwa Vidyapeetham
5 shared publicationsAbdulaziz Alhumam
4 shared publicationsRakesh Kumar
4 shared publications
Latest Publications
- 22 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
SmartSpice: A Flutter-Based Ensemble Deep Learning App for Real-Time Cardamom Early Leaf Disease Detection
SmartSpice presents a mobile foliar disease decision-support system using Flutter, which enables the detection of the early stages of foliar disease in cardamom, a high-value spice crop, which is severely hit by Leaf Blight and Leaf Spot that may result in yellowing and loss of yields by up to 40 percent. Conventional expert-based diagnosis is cumbersome, qualitative, and unavailable in remote plantations, and current AI solutions are generalized black boxes, which do not have cardamom-specific models, quantification of severity, or practical treatment advice, which reduces farmer trust and adoption. The suggested work utilizes a Dynamic Routing Capsule-Inspired Ensemble which is an integration of VGG-style CNN, ResNet50, and MobileNetV2 branches with adaptive input-specific weighting to allow healthy, blight, and leaf spot leaves to be classified effectively with multiscale feature extraction. The model is trained on a Kaggle curated dataset of 2,156 real-field images, which is augmented and stratified to split into two groups to better generalize and address the issue of class imbalance. Explainable AI with GradCAM also highlights the areas of lesions that underlie each prediction in a system, whereas computer-visionbased severity scores and severity-aware pesticide suggestions can also offer a more interpretable and accurate recommendation to farmers. It is established in the experiment that the ensemble is more accurate, precise, recalling, and has higher F1-score and ROC-AUC than standalone baselines (CNN, VGG16, VGG19, ResNet50) and can be inferred on a device. The resulting TensorFlow Lite model is embedded in a Flutter application supporting input via camera/gallery and entirely offline, real-time cardamom farmer diagnostics on resource-constrained smartphones, and thus SmartSpice is viable in remote areas of cardamom farmers.
- 22 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
SmartSpice: A Flutter-Based Ensemble Deep Learning App for Real-Time Cardamom Early Leaf Disease Detection
SmartSpice presents a mobile foliar disease decision-support system using Flutter, which enables the detection of the early stages of foliar disease in cardamom, a high-value spice crop, which is severely hit by Leaf Blight and Leaf Spot that may result in yellowing and loss of yields by up to 40 percent. Conventional expert-based diagnosis is cumbersome, qualitative, and unavailable in remote plantations, and current AI solutions are generalized black boxes, which do not have cardamom-specific models, quantification of severity, or practical treatment advice, which reduces farmer trust and adoption. The suggested work utilizes a Dynamic Routing Capsule-Inspired Ensemble which is an integration of VGG-style CNN, ResNet50, and MobileNetV2 branches with adaptive input-specific weighting to allow healthy, blight, and leaf spot leaves to be classified effectively with multiscale feature extraction. The model is trained on a Kaggle curated dataset of 2,156 real-field images, which is augmented and stratified to split into two groups to better generalize and address the issue of class imbalance. Explainable AI with GradCAM also highlights the areas of lesions that underlie each prediction in a system, whereas computer-visionbased severity scores and severity-aware pesticide suggestions can also offer a more interpretable and accurate recommendation to farmers. It is established in the experiment that the ensemble is more accurate, precise, recalling, and has higher F1-score and ROC-AUC than standalone baselines (CNN, VGG16, VGG19, ResNet50) and can be inferred on a device. The resulting TensorFlow Lite model is embedded in a Flutter application supporting input via camera/gallery and entirely offline, real-time cardamom farmer diagnostics on resource-constrained smartphones, and thus SmartSpice is viable in remote areas of cardamom farmers.
- 8 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Comparative Evaluation of Statistical Time-Series Models for Marine Heatwave Forecasting in the North Indian Ocean
Over the last forty years, there is an increasing rate in the occurrence of intense oceanic heatwaves in the Indian Ocean. Marine heatwaves are extremely dangerous to the ecosystems, climatic conditions, and the economies. Sea surface temperature (SST) anomalies also need to be forecasted accurately so as to be able to detect and monitor these events at an early stage. This paper uses the OISST data to create an all-encompassing statistical forecasting model of Indian Ocean SST anomalies. The work considers seven models, ARIMA, SARIMA, Holt-Winters, Seasonal Exponential Smoothing, Theta, STL+ETS, and Prophet, which are applied and tested on the use of anomaly time-series based on 1981-2025. STL+ETS turned out to be the most productive model with the smallest errors (MAE: 0.134-0.161, MSE: 0.032-0.041, RMSE: 0.179–0.203). The results indicate the potential high usefulness of classical statistical models in operational MHW early-warning applications and provide a basis on how statistical forecasting may be used together with machine-learning and hybrid ocean prediction systems
- 8 April 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Comparative Evaluation of Statistical Time-Series Models for Marine Heatwave Forecasting in the North Indian Ocean
Over the last forty years, there is an increasing rate in the occurrence of intense oceanic heatwaves in the Indian Ocean. Marine heatwaves are extremely dangerous to the ecosystems, climatic conditions, and the economies. Sea surface temperature (SST) anomalies also need to be forecasted accurately so as to be able to detect and monitor these events at an early stage. This paper uses the OISST data to create an all-encompassing statistical forecasting model of Indian Ocean SST anomalies. The work considers seven models, ARIMA, SARIMA, Holt-Winters, Seasonal Exponential Smoothing, Theta, STL+ETS, and Prophet, which are applied and tested on the use of anomaly time-series based on 1981-2025. STL+ETS turned out to be the most productive model with the smallest errors (MAE: 0.134-0.161, MSE: 0.032-0.041, RMSE: 0.179–0.203). The results indicate the potential high usefulness of classical statistical models in operational MHW early-warning applications and provide a basis on how statistical forecasting may be used together with machine-learning and hybrid ocean prediction systems
- Open Access
- 30 March 2026
- Published by Springer Nature in Discover Computing
- Vol. 29 (1)
Adaptive stock price volatility forecasting using enhanced kMedoids clustering with attention based LSTM for day trading
- 30 March 2026
- Published by Springer Nature in Discover Computing
- Vol. 29 (1)
Open AccessAdaptive stock price volatility forecasting using enhanced kMedoids clustering with attention based LSTM for day trading
- Open Access
- 28 January 2026
- Published by Springer Nature in Discover Computing
- Vol. 29 (1) , 53
Privacy preserving epileptic seizure recognition using federated and explainable machine learning
- 28 January 2026
- Published by Springer Nature in Discover Computing
- Vol. 29 (1) , 53
Open AccessPrivacy preserving epileptic seizure recognition using federated and explainable machine learning
- Top Cited PapersOpen Access
- 28 January 2026
- Published by Frontiers Media SA in Frontiers in Medicine
- Vol. 12
Exploring Agentic AI in Healthcare: A Study on Its Working Mechanism
Introduction: Rapid advancements in artificial intelligence (AI) have ushered in an era of hyperautomation and intelligent orchestration across multiple engineering domains, with healthcare emerging as one of the most impactful application areas. Among recent developments, Agentic AI has gained attention as a sub-domain of AI capable of autonomous operation, decision-making, and goal-driven behavior with minimal human intervention. This study aims to explore the architectural and functional role of Agentic AI in modern healthcare systems. Methods: The study adopts a conceptual and analytical approach to examine the core components of Agentic AI, including agent design, decision-making mechanisms, task allocation strategies, agent coordination, and ranking frameworks. It further investigates the integration of emerging 6G networking technologies within Agentic AI architectures. A qualitative case study on remote robotic surgery is presented to illustrate practical applicability. Additionally, a Strengths, Weaknesses, Opportunities, and Threats (SWOT) analysis is conducted to assess strategic and operational considerations. Results: The analysis demonstrates that Agentic AI architectures, when supported by high-speed and low-latency 6G communication, can enable efficient autonomous decision-making and coordinated task execution in complex healthcare workflows. The case study highlights the feasibility of Agentic AI in enabling remote robotic surgery with enhanced responsiveness, precision, and reliability. The SWOT analysis reveals strong potential for scalability and efficiency while also identifying challenges related to ethical governance, system robustness, and security. Discussion: The findings suggest that Agentic AI represents a promising paradigm for next-generation healthcare systems, particularly in remote and critical care applications. While the proposed framework offers architectural insights and strategic value, responsible integration requires addressing limitations such as trust, regulatory compliance, and system transparency. Overall, this study provides a holistic understanding of how Agentic AI can be effectively and ethically integrated into healthcare ecosystems.
- 28 January 2026
- Published by Frontiers Media SA in Frontiers in Medicine
- Vol. 12
Top Cited PapersOpen AccessExploring Agentic AI in Healthcare: A Study on Its Working Mechanism
Introduction: Rapid advancements in artificial intelligence (AI) have ushered in an era of hyperautomation and intelligent orchestration across multiple engineering domains, with healthcare emerging as one of the most impactful application areas. Among recent developments, Agentic AI has gained attention as a sub-domain of AI capable of autonomous operation, decision-making, and goal-driven behavior with minimal human intervention. This study aims to explore the architectural and functional role of Agentic AI in modern healthcare systems. Methods: The study adopts a conceptual and analytical approach to examine the core components of Agentic AI, including agent design, decision-making mechanisms, task allocation strategies, agent coordination, and ranking frameworks. It further investigates the integration of emerging 6G networking technologies within Agentic AI architectures. A qualitative case study on remote robotic surgery is presented to illustrate practical applicability. Additionally, a Strengths, Weaknesses, Opportunities, and Threats (SWOT) analysis is conducted to assess strategic and operational considerations. Results: The analysis demonstrates that Agentic AI architectures, when supported by high-speed and low-latency 6G communication, can enable efficient autonomous decision-making and coordinated task execution in complex healthcare workflows. The case study highlights the feasibility of Agentic AI in enabling remote robotic surgery with enhanced responsiveness, precision, and reliability. The SWOT analysis reveals strong potential for scalability and efficiency while also identifying challenges related to ethical governance, system robustness, and security. Discussion: The findings suggest that Agentic AI represents a promising paradigm for next-generation healthcare systems, particularly in remote and critical care applications. While the proposed framework offers architectural insights and strategic value, responsible integration requires addressing limitations such as trust, regulatory compliance, and system transparency. Overall, this study provides a holistic understanding of how Agentic AI can be effectively and ethically integrated into healthcare ecosystems.
- Open Access
- 14 January 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Optimizing sepsis mortality prediction using hybrid federated learning and explainable AI framework
Sepsis is a life-threatening condition resulting from a dysregulated host response to infection, frequently leading to organ failure and high mortality in hospital settings. Early identification of sepsis is critical for reducing mortality; however, conventional diagnostic approaches often fail to capture complex clinical patterns at an early stage. Recent advances in machine learning (ML) and explainable artificial intelligence (XAI) have demonstrated potential for improving predictive accuracy while supporting clinical interpretability. Nevertheless, concerns related to data privacy and model transparency continue to limit real-world clinical adoption. To address these challenges, this study proposes a hybrid framework that integrates federated learning with ensemble-based machine learning models and explainable AI techniques for sepsis mortality prediction. The framework employs Random Forest, LightGBM, XGBoost, K-Nearest Neighbors, and Logistic Regression models, trained in a decentralized manner to preserve patient data privacy. Model interpretability is enhanced using SHapley Additive exPlanations (SHAP), Local Interpretable Model-agnostic Explanations (LIME), and Partial Dependence Plots (PDP), enabling transparent and clinician-oriented decision support. The proposed framework is evaluated using standard performance metrics, including accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (ROC–AUC), in both centralized and federated settings. Experimental results demonstrate that ensemble models, particularly Random Forest and gradient boosting methods, achieve high predictive performance while maintaining robustness in a federated environment. The findings indicate that combining FL with XAI enables accurate, privacy-preserving, and interpretable sepsis mortality prediction, supporting reliable clinical decision-making and potential deployment in real-time intensive care unit applications.
- 14 January 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessOptimizing sepsis mortality prediction using hybrid federated learning and explainable AI framework
Sepsis is a life-threatening condition resulting from a dysregulated host response to infection, frequently leading to organ failure and high mortality in hospital settings. Early identification of sepsis is critical for reducing mortality; however, conventional diagnostic approaches often fail to capture complex clinical patterns at an early stage. Recent advances in machine learning (ML) and explainable artificial intelligence (XAI) have demonstrated potential for improving predictive accuracy while supporting clinical interpretability. Nevertheless, concerns related to data privacy and model transparency continue to limit real-world clinical adoption. To address these challenges, this study proposes a hybrid framework that integrates federated learning with ensemble-based machine learning models and explainable AI techniques for sepsis mortality prediction. The framework employs Random Forest, LightGBM, XGBoost, K-Nearest Neighbors, and Logistic Regression models, trained in a decentralized manner to preserve patient data privacy. Model interpretability is enhanced using SHapley Additive exPlanations (SHAP), Local Interpretable Model-agnostic Explanations (LIME), and Partial Dependence Plots (PDP), enabling transparent and clinician-oriented decision support. The proposed framework is evaluated using standard performance metrics, including accuracy, precision, recall, F1-score, and area under the receiver operating characteristic curve (ROC–AUC), in both centralized and federated settings. Experimental results demonstrate that ensemble models, particularly Random Forest and gradient boosting methods, achieve high predictive performance while maintaining robustness in a federated environment. The findings indicate that combining FL with XAI enables accurate, privacy-preserving, and interpretable sepsis mortality prediction, supporting reliable clinical decision-making and potential deployment in real-time intensive care unit applications.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
CottonCare: A Flutter-Based Mobile Application Leveraging Hybrid ResNet50-ViT Architecture for Enhanced Cotton Leaf Disease Detection
The early detection of plant diseases helps promote precision farming and higher yield, thereby improving farmers ’ income and supporting sustainable agricultural practices. In particular, early detection of cotton leaf disease ensures better fiber quality. This research presents an intelligent deep learning-based diagnostic system that leverages a hybrid architecture combining ResNet50 and Vision Transformers (ViT) to accurately indicate the type of cotton leaf disease to farmers without agronomy expertise. The data set undergoes a structured preprocessing pipeline that includes resizing, normalization, color space standardization, and array conversion. The proposed hybrid model (ResNet50 + ViT) was trained, evaluated, and compared against deep learning models, such as CNN, VGG16, Inception, ResNet152V2, ResNet50, DenseNet121, Xception, MobileNetV2, and Random Forest. Among these, the hybrid model demonstrated the best performance, achieving an accuracy: 99%, precision: 98%, recall: 97%, and F1-score: 98%. The trained model was seamlessly deployed into a Flutter-based mobile application, allowing users to detect diseases in real-time from captured leaf images. Our proposed work highlights the potential of combining CNNs with transformer-based architecture for discerning cotton foliage disease, providing farmers with an AI-powered mobile diagnostic tool for field-level decision-making.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
CottonCare: A Flutter-Based Mobile Application Leveraging Hybrid ResNet50-ViT Architecture for Enhanced Cotton Leaf Disease Detection
The early detection of plant diseases helps promote precision farming and higher yield, thereby improving farmers ’ income and supporting sustainable agricultural practices. In particular, early detection of cotton leaf disease ensures better fiber quality. This research presents an intelligent deep learning-based diagnostic system that leverages a hybrid architecture combining ResNet50 and Vision Transformers (ViT) to accurately indicate the type of cotton leaf disease to farmers without agronomy expertise. The data set undergoes a structured preprocessing pipeline that includes resizing, normalization, color space standardization, and array conversion. The proposed hybrid model (ResNet50 + ViT) was trained, evaluated, and compared against deep learning models, such as CNN, VGG16, Inception, ResNet152V2, ResNet50, DenseNet121, Xception, MobileNetV2, and Random Forest. Among these, the hybrid model demonstrated the best performance, achieving an accuracy: 99%, precision: 98%, recall: 97%, and F1-score: 98%. The trained model was seamlessly deployed into a Flutter-based mobile application, allowing users to detect diseases in real-time from captured leaf images. Our proposed work highlights the potential of combining CNNs with transformer-based architecture for discerning cotton foliage disease, providing farmers with an AI-powered mobile diagnostic tool for field-level decision-making.
- Open Access
- 9 October 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-21
IoMT driven Alzheimer’s prediction model empowered with transfer learning and explainable AI approach in healthcare 5.0
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the primary cause of dementia, responsible for 60–70% of global cases. It severely affects memory, cognitive function, and daily independence, placing a substantial emotional and economic burden on patients and caregivers. Early and accurate prediction remains difficult due to the high cost of neuroimaging, scarcity of annotated datasets, and the “black-box” nature of most artificial intelligence (AI) models. With the emergence of Healthcare 5.0, the Internet of Medical Things (IoMT) offers new opportunities for patient-centric, real-time monitoring and data-driven diagnosis. This study proposes an IoMT-driven Alzheimer’s prediction framework that combines transfer learning (ResNet152) with explainable AI (XAI) to provide both accuracy and interpretability. The publicly available Kaggle Alzheimer’s MRI dataset, comprising 33,984 images across four classes (Non-Demented, Very Mild, Mild, and Moderate Demented) was employed. To address class imbalance, a Conditional Wasserstein GAN was applied for synthetic image generation and balanced sampling. The proposed ResNet152-TL-XAI model achieved 97.77% accuracy, with a precision of 0.981, recall of 0.987, F1-score of 0.983, and specificity of 99.13%, outperforming several state-of-the-art methods. Interpretability was ensured through Grad-CAM, SHAP, and LIME, which consistently highlighted clinically relevant brain regions such as the Hippocampus and ventricles, confirming biological plausibility and increasing clinician trust. By integrating IoMT-enabled data acquisition, transfer learning for efficient training, and multi-method XAI for transparency, the proposed pipeline demonstrates strong potential for early, accurate, and interpretable Alzheimer’s staging. These results position the framework as a practical candidate for integration into Healthcare 5.0 ecosystems, supporting timely diagnosis, patient monitoring, and personalized interventions.
- 9 October 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-21
Open AccessIoMT driven Alzheimer’s prediction model empowered with transfer learning and explainable AI approach in healthcare 5.0
Alzheimer’s disease (AD) is a progressive neurodegenerative disorder and the primary cause of dementia, responsible for 60–70% of global cases. It severely affects memory, cognitive function, and daily independence, placing a substantial emotional and economic burden on patients and caregivers. Early and accurate prediction remains difficult due to the high cost of neuroimaging, scarcity of annotated datasets, and the “black-box” nature of most artificial intelligence (AI) models. With the emergence of Healthcare 5.0, the Internet of Medical Things (IoMT) offers new opportunities for patient-centric, real-time monitoring and data-driven diagnosis. This study proposes an IoMT-driven Alzheimer’s prediction framework that combines transfer learning (ResNet152) with explainable AI (XAI) to provide both accuracy and interpretability. The publicly available Kaggle Alzheimer’s MRI dataset, comprising 33,984 images across four classes (Non-Demented, Very Mild, Mild, and Moderate Demented) was employed. To address class imbalance, a Conditional Wasserstein GAN was applied for synthetic image generation and balanced sampling. The proposed ResNet152-TL-XAI model achieved 97.77% accuracy, with a precision of 0.981, recall of 0.987, F1-score of 0.983, and specificity of 99.13%, outperforming several state-of-the-art methods. Interpretability was ensured through Grad-CAM, SHAP, and LIME, which consistently highlighted clinically relevant brain regions such as the Hippocampus and ventricles, confirming biological plausibility and increasing clinician trust. By integrating IoMT-enabled data acquisition, transfer learning for efficient training, and multi-method XAI for transparency, the proposed pipeline demonstrates strong potential for early, accurate, and interpretable Alzheimer’s staging. These results position the framework as a practical candidate for integration into Healthcare 5.0 ecosystems, supporting timely diagnosis, patient monitoring, and personalized interventions.
- 18 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
A Scalable Serverless Framework for Real-Time Analytics of Remote Patient Monitoring IoT Devices
The exponential growth of IoT-equipped remote patient monitoring (RPM) systems has generated enormous amounts of real-time health data with immense challenges related to latency, scalability, and cost of operation. Conventional cloud-based analytical platforms tend to be plagued with high overheads, sluggish response times, and inflexibility, rendering them inappropriate for critical healthcare applications. This research puts forward (SHEAF) Serverless Health Edge Analytics Framework, a light, scalable, and affordable Serverless architecture that is tailored for real-time data processing and analysis of health information from RPM IoT devices. The architecture uses MQTT for communication at the device level, Kafka for high-volume data ingestion, and AWS Lambda as the Serverless compute engine to analyze data streams in real time. On-site filtering and simple anomaly detection are done by an optional TinyML module embedded on edge devices, reducing transmission burden. Predictive analytics with light LSTM models are embedded in Serverless functions to identify anomalies and trigger alerts. Based on simulated trials with synthetic patient data, they exhibited a reduction of 40–60% in latency, 92–95% correct prediction of the occurrence of important events, and a 30% reduction of bandwidth consumption versus standard cloud systems. Moreover, the system held consistent performance using more than 1000 connections simultaneously on a device, confirming good scalability. Its Serverless core presents a low-latency, low-cost, and high-performance proposition against conventional frameworks, making it particularly well suited for under provisioned or rural health networks.
- 18 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
A Scalable Serverless Framework for Real-Time Analytics of Remote Patient Monitoring IoT Devices
The exponential growth of IoT-equipped remote patient monitoring (RPM) systems has generated enormous amounts of real-time health data with immense challenges related to latency, scalability, and cost of operation. Conventional cloud-based analytical platforms tend to be plagued with high overheads, sluggish response times, and inflexibility, rendering them inappropriate for critical healthcare applications. This research puts forward (SHEAF) Serverless Health Edge Analytics Framework, a light, scalable, and affordable Serverless architecture that is tailored for real-time data processing and analysis of health information from RPM IoT devices. The architecture uses MQTT for communication at the device level, Kafka for high-volume data ingestion, and AWS Lambda as the Serverless compute engine to analyze data streams in real time. On-site filtering and simple anomaly detection are done by an optional TinyML module embedded on edge devices, reducing transmission burden. Predictive analytics with light LSTM models are embedded in Serverless functions to identify anomalies and trigger alerts. Based on simulated trials with synthetic patient data, they exhibited a reduction of 40–60% in latency, 92–95% correct prediction of the occurrence of important events, and a 30% reduction of bandwidth consumption versus standard cloud systems. Moreover, the system held consistent performance using more than 1000 connections simultaneously on a device, confirming good scalability. Its Serverless core presents a low-latency, low-cost, and high-performance proposition against conventional frameworks, making it particularly well suited for under provisioned or rural health networks.
- Open Access
- 18 July 2025
- Published by Elsevier in Informatics in Medicine Unlocked
- Vol. 57
A novel hybrid model for brain tumor analysis with CNN and Moth Flame Optimization
- 18 July 2025
- Published by Elsevier in Informatics in Medicine Unlocked
- Vol. 57
Open AccessA novel hybrid model for brain tumor analysis with CNN and Moth Flame Optimization