Deepak Dudeja
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Deepak Dudeja has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
Gunjan Chhabra
3 shared publicationsLalit Garg
3 shared publicationsVijay Prakash
3 shared publicationsAnil Garg
Maharishi Markandeshwar Engineering College, MM(DU),Department of Electronics and Communication Engineering,Mullana-Ambala,Haryana,India,133207
2 shared publicationsB K Agrawal
2 shared publicationsDeepika Arora
2 shared publications
Latest Publications
- 1 February 2026
- Published by Science Publications in Journal of Computer Science
- Vol. 22 (2) , 738-746
An Intelligent System for Cardiovascular Disease Prediction Using Deep Learning
- 1 February 2026
- Published by Science Publications in Journal of Computer Science
- Vol. 22 (2) , 738-746
An Intelligent System for Cardiovascular Disease Prediction Using Deep Learning
- 10 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Prediction Systems in Angina Pectoris Monitoring: A Review
AI-ML platforms have the potential to have a significant influence on medicine in the future. In reality, AI-powered diagnostics will increase precision and efficiency, which will eventually result in better patient outcomes and lower medical expenses. Machine learning is a promising solution to improve forecast accuracy since small risk indicators may not always be detected by traditional diagnostic techniques. With differing degrees of effectiveness, a number of machine learning (ML) algorithms have been used to the prediction of cardiac disease. Commonly utilized Random Forest, Decision Trees, and Support Vector Machines (SVMs) have demonstrated encouraging outcomes, with some research achieving accuracies of up to 97%. Other methods have also been used, such as k-Nearest Neighbors (KNN), Logistic Regression, and Artificial Neural Networks (ANNs).
- 10 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Prediction Systems in Angina Pectoris Monitoring: A Review
AI-ML platforms have the potential to have a significant influence on medicine in the future. In reality, AI-powered diagnostics will increase precision and efficiency, which will eventually result in better patient outcomes and lower medical expenses. Machine learning is a promising solution to improve forecast accuracy since small risk indicators may not always be detected by traditional diagnostic techniques. With differing degrees of effectiveness, a number of machine learning (ML) algorithms have been used to the prediction of cardiac disease. Commonly utilized Random Forest, Decision Trees, and Support Vector Machines (SVMs) have demonstrated encouraging outcomes, with some research achieving accuracies of up to 97%. Other methods have also been used, such as k-Nearest Neighbors (KNN), Logistic Regression, and Artificial Neural Networks (ANNs).
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Towards Intelligent Anemia Detection: An Empirical Analysis of Clinical and Imaging-Based Machine Learning Approaches
Anemia affects more than 1.6 billion people worldwide and remains a critical global health concern due to its impact on oxygen delivery and patient well-being. Conventional diagnostic methods such as complete blood count (CBC) and peripheral blood smear (PBS) analysis, while reliable, are labor-intensive, resource-dependent, and impractical for large-scale screening in low-resource settings. Recent advances in artificial intelligence (AI) and deep learning (DL) provide promising opportunities for automated, accurate, and scalable anemia detection. Unlike prior works that focus solely on reviewing techniques or applying traditional ML models with handcrafted features, this paper makes a dual contribution: (1) it presents a systematic review of machine learning and deep learning approaches for anemia detection, highlighting the evolution from conventional classifiers to CNNs, Vision Transformers, and hybrid multimodal frameworks; and (2) it introduces and empirically validates a fine-tuned ResNet50 model applied to PBS images. The proposed framework achieves state-of-the-art performance with 98% accuracy and a 99.1% F1-score, outperforming several existing methods. In addition, this work identifies key challenges—including dataset scarcity, generalization across diverse imaging conditions, and model interpretability—and outlines future directions such as multimodal fusion, explainable AI, and clinical integration. Together, these contributions advance the development of intelligent and clinically viable anemia detection systems for improved healthcare outcomes.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Towards Intelligent Anemia Detection: An Empirical Analysis of Clinical and Imaging-Based Machine Learning Approaches
Anemia affects more than 1.6 billion people worldwide and remains a critical global health concern due to its impact on oxygen delivery and patient well-being. Conventional diagnostic methods such as complete blood count (CBC) and peripheral blood smear (PBS) analysis, while reliable, are labor-intensive, resource-dependent, and impractical for large-scale screening in low-resource settings. Recent advances in artificial intelligence (AI) and deep learning (DL) provide promising opportunities for automated, accurate, and scalable anemia detection. Unlike prior works that focus solely on reviewing techniques or applying traditional ML models with handcrafted features, this paper makes a dual contribution: (1) it presents a systematic review of machine learning and deep learning approaches for anemia detection, highlighting the evolution from conventional classifiers to CNNs, Vision Transformers, and hybrid multimodal frameworks; and (2) it introduces and empirically validates a fine-tuned ResNet50 model applied to PBS images. The proposed framework achieves state-of-the-art performance with 98% accuracy and a 99.1% F1-score, outperforming several existing methods. In addition, this work identifies key challenges—including dataset scarcity, generalization across diverse imaging conditions, and model interpretability—and outlines future directions such as multimodal fusion, explainable AI, and clinical integration. Together, these contributions advance the development of intelligent and clinically viable anemia detection systems for improved healthcare outcomes.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Deep Learning in Dermatology: An Empirical Analysis of CNN Architectures and Diagnostic Accuracy
Skin cancer is exceedingly prevalent globally and melanoma and non-melanoma are significant reasons for morbidity and mortality. A brief delay in discovering breast cancer puts it at higher risk of spreading and becoming more difficult to treat. Visual check and microscope examination, today's primary modes of diagnosis, are time-consuming, invasive and have potential for errors and variability among experts. Although dermoscopy enables skin lesions to be visualized more easily, its performance depends a great deal on how well dermatologists can differentiate similar-appearing lesion types. Therefore, automated, straightforward and trustworthy tools that support doctors in the early detection of cancer are in persistent demand. Deep learning (DL) has emerged as an important technology in the analysis of medical images, working extremely well at segmentation, classification and detecting abnormalities.
- 28 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Deep Learning in Dermatology: An Empirical Analysis of CNN Architectures and Diagnostic Accuracy
Skin cancer is exceedingly prevalent globally and melanoma and non-melanoma are significant reasons for morbidity and mortality. A brief delay in discovering breast cancer puts it at higher risk of spreading and becoming more difficult to treat. Visual check and microscope examination, today's primary modes of diagnosis, are time-consuming, invasive and have potential for errors and variability among experts. Although dermoscopy enables skin lesions to be visualized more easily, its performance depends a great deal on how well dermatologists can differentiate similar-appearing lesion types. Therefore, automated, straightforward and trustworthy tools that support doctors in the early detection of cancer are in persistent demand. Deep learning (DL) has emerged as an important technology in the analysis of medical images, working extremely well at segmentation, classification and detecting abnormalities.
- 30 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 177-183
Integrating AI in Drone Systems: Advancements, Challenges, and Future Directions
Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) technologies-based robotic drones have brought a revolution primarily in the business of agriculture, surveillance, disaster management, and logistics. This paper tries to deliver a comprehension into these advancements that fundamentally prove to be beneficial in upgrading robotic drones. AI lists some of the abilities that it imparts to the drones so that they can accomplish more complicated tasks like navigation, obstacle avoidance and decision making. ML algorithms allow the drone to learn from experience i.e., data improving with time its performance in such activities as object detection path planning predictive maintenance etc. DL, a smaller division of ML, uses the concept of neural networks to work on a large volume of data and hence accomplish such tasks requiring great accuracy as image and speech recognition. This paper presents an overview based on the main research trends and technical approaches in addition to applications at hand for AI, ML, and DL within robotic drones. It further identifies possible issues that could exist in this field including computational challenges, data privacy implications, and algorithms that need to be strong toward dynamic environments. Some mathematical representations have been made to show both the complexities and solutions that exist within the technology of drones. Also, the review points out key materials and methods that tackle these issues, highlighting the value of cross-discipline study and teamwork. The paper wraps up with a summary of the findings and suggests directions for future research to overcome current challenges as well as improve the productivity and effectiveness of robotic drones. This detailed review tries to give a better view of the current situation and future outlook of AI, ML, and DL in robotic drones while giving helpful ideas to researchers, experts, and decision-makers.
- 30 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 177-183
Integrating AI in Drone Systems: Advancements, Challenges, and Future Directions
Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL) technologies-based robotic drones have brought a revolution primarily in the business of agriculture, surveillance, disaster management, and logistics. This paper tries to deliver a comprehension into these advancements that fundamentally prove to be beneficial in upgrading robotic drones. AI lists some of the abilities that it imparts to the drones so that they can accomplish more complicated tasks like navigation, obstacle avoidance and decision making. ML algorithms allow the drone to learn from experience i.e., data improving with time its performance in such activities as object detection path planning predictive maintenance etc. DL, a smaller division of ML, uses the concept of neural networks to work on a large volume of data and hence accomplish such tasks requiring great accuracy as image and speech recognition. This paper presents an overview based on the main research trends and technical approaches in addition to applications at hand for AI, ML, and DL within robotic drones. It further identifies possible issues that could exist in this field including computational challenges, data privacy implications, and algorithms that need to be strong toward dynamic environments. Some mathematical representations have been made to show both the complexities and solutions that exist within the technology of drones. Also, the review points out key materials and methods that tackle these issues, highlighting the value of cross-discipline study and teamwork. The paper wraps up with a summary of the findings and suggests directions for future research to overcome current challenges as well as improve the productivity and effectiveness of robotic drones. This detailed review tries to give a better view of the current situation and future outlook of AI, ML, and DL in robotic drones while giving helpful ideas to researchers, experts, and decision-makers.
- 8 June 2025
- Published by Springer Nature
- p. 67-82
Wind Turbine Reliability Through AI-Driven Predictive Maintenance and IIoT Integration
- 8 June 2025
- Published by Springer Nature
- p. 67-82
Wind Turbine Reliability Through AI-Driven Predictive Maintenance and IIoT Integration
- 31 March 2025
- Published by Taylor & Francis
- p. 527-533
Android offloading computation over the cloud
This paper analyzes whether Android apps may outsource computational activities to cloud servers. Due to the complexity of mobile apps, shifting computing operations to the cloud has become popular to improve performance and reduce energy consumption. This paper explains how to deploy an Android app on Cloudflare and tests it under various settings. The testing showed that cloud computing can boost Android app efficiency. The findings show that shifting computing workloads to the cloud reduces mobile device CPU utilization. This frees up device resources, allowing it to perform other tasks. However, cloud processing has limitations. Network latency between the mobile device and the cloud server may affect application performance. In addition, the cost of offloading computation to the cloud can be a concern for some users, especially in high data consumption scenarios. Our findings can be helpful for developers who want to optimize the performance of their Android applications and for researchers who wish to investigate this topic further. Future research could explore ways to minimize the impact of network latency on offloading computation and investigate the cost-effectiveness of this technique for different types of applications.
- 31 March 2025
- Published by Taylor & Francis
- p. 527-533
Android offloading computation over the cloud
This paper analyzes whether Android apps may outsource computational activities to cloud servers. Due to the complexity of mobile apps, shifting computing operations to the cloud has become popular to improve performance and reduce energy consumption. This paper explains how to deploy an Android app on Cloudflare and tests it under various settings. The testing showed that cloud computing can boost Android app efficiency. The findings show that shifting computing workloads to the cloud reduces mobile device CPU utilization. This frees up device resources, allowing it to perform other tasks. However, cloud processing has limitations. Network latency between the mobile device and the cloud server may affect application performance. In addition, the cost of offloading computation to the cloud can be a concern for some users, especially in high data consumption scenarios. Our findings can be helpful for developers who want to optimize the performance of their Android applications and for researchers who wish to investigate this topic further. Future research could explore ways to minimize the impact of network latency on offloading computation and investigate the cost-effectiveness of this technique for different types of applications.
- 31 March 2025
- Published by Taylor & Francis
- p. 519-526
Encryption and decryption framework using elliptic curve cryptography
Secure file encryption has become increasingly important in various disciplines, including finance, healthcare, and government, where protecting sensitive data is paramount. The proposed framework aims to meet this requirement by developing an encryption and decryption framework for files that employ Elliptic Curve Cryptography (ECC), a modern and practical cryptographic technique. The framework will employ ECC algorithms for key generation, encryption, and decryption, which offer several advantages over conventional methods such as RSA and AES. ECC provides equivalent security with shortened key lengths, resulting in faster encryption and decryption, decreased computational burden, and optimized resource utilization. A crucial aspect of the proposed framework is its user-friendly interface, making it easy for users to manage encrypted files. In addition, the framework will include key management and validation mechanisms to ensure the security and authenticity of encryption keys, thereby preventing unauthorized access to encrypted files. The proposed framework will be rigorously tested on a local machine and hosted on a server in the cloud to ensure its dependability and resiliency in real-world scenarios. The framework's efficacy will be determined based on the encryption and decryption times for various file sizes. The outcomes will demonstrate the efficacy and efficiency of the proposed framework, demonstrating that it provides superior performance compared to conventional algorithms.
- 31 March 2025
- Published by Taylor & Francis
- p. 519-526
Encryption and decryption framework using elliptic curve cryptography
Secure file encryption has become increasingly important in various disciplines, including finance, healthcare, and government, where protecting sensitive data is paramount. The proposed framework aims to meet this requirement by developing an encryption and decryption framework for files that employ Elliptic Curve Cryptography (ECC), a modern and practical cryptographic technique. The framework will employ ECC algorithms for key generation, encryption, and decryption, which offer several advantages over conventional methods such as RSA and AES. ECC provides equivalent security with shortened key lengths, resulting in faster encryption and decryption, decreased computational burden, and optimized resource utilization. A crucial aspect of the proposed framework is its user-friendly interface, making it easy for users to manage encrypted files. In addition, the framework will include key management and validation mechanisms to ensure the security and authenticity of encryption keys, thereby preventing unauthorized access to encrypted files. The proposed framework will be rigorously tested on a local machine and hosted on a server in the cloud to ensure its dependability and resiliency in real-world scenarios. The framework's efficacy will be determined based on the encryption and decryption times for various file sizes. The outcomes will demonstrate the efficacy and efficiency of the proposed framework, demonstrating that it provides superior performance compared to conventional algorithms.
- 31 March 2025
- Published by Taylor & Francis
- p. 534-545
Cryptography machine: Encryption and decryption model for multimedia data
The digital era has brought a surge in the amount of data generated, increasing the need for data security across individuals, organizations, and governments. Protecting sensitive information from unauthorized access is a top priority in today's digital world. Cryptography is used to ensure secure communication over the internet, including email, instant messaging, and web browsing. Our paper's primary goal is to develop a cryptography machine capable of encrypting and decrypting different data types, such as plain text, images, audio and videos using the cryptography of the symmetric key. The proposed model will be implemented using Python and Java. Developing the cryptography machine will significantly protect sensitive information and ensure its confidentiality and integrity. The machine's ability to encrypt and decrypt different data types using symmetric-key cryptography will provide an added layer of security to individuals, organizations, and governments in the digital era.
- 31 March 2025
- Published by Taylor & Francis
- p. 534-545
Cryptography machine: Encryption and decryption model for multimedia data
The digital era has brought a surge in the amount of data generated, increasing the need for data security across individuals, organizations, and governments. Protecting sensitive information from unauthorized access is a top priority in today's digital world. Cryptography is used to ensure secure communication over the internet, including email, instant messaging, and web browsing. Our paper's primary goal is to develop a cryptography machine capable of encrypting and decrypting different data types, such as plain text, images, audio and videos using the cryptography of the symmetric key. The proposed model will be implemented using Python and Java. Developing the cryptography machine will significantly protect sensitive information and ensure its confidentiality and integrity. The machine's ability to encrypt and decrypt different data types using symmetric-key cryptography will provide an added layer of security to individuals, organizations, and governments in the digital era.
- 1 January 2025
- Published by Taru Publications in Journal of Discrete Mathematical Sciences and Cryptography
- Vol. 28 (8) , 3001-3011
Enhancing cybersecurity measures in virtual reality and augmented reality environments : Challenges, risks, and solutions
New opportunities for immersive and interactive experiences have been created by the incorporation of Virtual Reality (VR) and Augmented Reality (AR) technology into a variety of industries, including business, healthcare, education, and entertainment. But this progress also brings with it hitherto unheard-of cybersecurity risks. Cyberattacks are drawn to VR/AR environments because of the complex data streams they include, including biometric data, personal data, and real-time geographical mapping. This study investigates the cybersecurity threats related to VR/AR platforms, including identity theft, device manipulation, data breaches, and invasions of privacy. It also covers new dangers such as malware directed at immersive apps and the use of sensory weaknesses to create bodily damage in real life. These vulnerabilities are made worse by the absence of standardized security standards in VR/AR systems, underscoring the necessity of strong security frameworks. End-to-end encryption, multi-factor authentication, biometric data protection, and the creation of safe software and hardware ecosystems are highlighted as viable remedies. In addition, ethical issues pertaining to data ownership and privacy are explored, along with suggestions for legislative actions to shield consumers from online dangers. Maintaining the security of VR/AR technologies is essential to building confidence and providing consumers worldwide with secure, immersive experiences as they develop.
- 1 January 2025
- Published by Taru Publications in Journal of Discrete Mathematical Sciences and Cryptography
- Vol. 28 (8) , 3001-3011
Enhancing cybersecurity measures in virtual reality and augmented reality environments : Challenges, risks, and solutions
New opportunities for immersive and interactive experiences have been created by the incorporation of Virtual Reality (VR) and Augmented Reality (AR) technology into a variety of industries, including business, healthcare, education, and entertainment. But this progress also brings with it hitherto unheard-of cybersecurity risks. Cyberattacks are drawn to VR/AR environments because of the complex data streams they include, including biometric data, personal data, and real-time geographical mapping. This study investigates the cybersecurity threats related to VR/AR platforms, including identity theft, device manipulation, data breaches, and invasions of privacy. It also covers new dangers such as malware directed at immersive apps and the use of sensory weaknesses to create bodily damage in real life. These vulnerabilities are made worse by the absence of standardized security standards in VR/AR systems, underscoring the necessity of strong security frameworks. End-to-end encryption, multi-factor authentication, biometric data protection, and the creation of safe software and hardware ecosystems are highlighted as viable remedies. In addition, ethical issues pertaining to data ownership and privacy are explored, along with suggestions for legislative actions to shield consumers from online dangers. Maintaining the security of VR/AR technologies is essential to building confidence and providing consumers worldwide with secure, immersive experiences as they develop.