EXPLAINABLE AND HUMAN-CENTRIC AI FOR CYBER-PHYSICAL AND HEALTHCARE SECURITY

Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [123] EXPLAINABLE AND HUMAN-CENTRIC AI FOR CYBER-PHYSICAL AND HEALTHCARE SECURITY Yashwant Reddy, M.Sc. CSE., Jawaharlal Nehru University (JNU), Delhi. [email protected] Priya Mehta, M.Sc. CSE., Jawaharlal Nehru University (JNU), Delhi. [email protected] Mitali Deshmukh, M.Sc. CSE., Jawaharlal Nehru University (JNU), Delhi. [email protected] ABSTRACT The increasing reliance on cyber-physical systems and healthcare technologies has heightened the risks associated with cybersecurity threats and data breaches. Traditional AI models, while effective, often operate as "black boxes," making them difficult to trust and interpret, particularly in critical sectors like healthcare and cyber-physical systems. This paper explores the development and application of explainable AI (XAI) in the context of cyberphysical and healthcare security. By focusing on transparency, accountability, and human-centered design, the proposed XAI frameworks aim to bridge the gap between machine learning algorithms and human decision-makers. The integration of XAI ensures that AI-driven security systems are not only accurate but also interpretable, enabling security personnel and healthcare professionals to understand, trust, and effectively interact with the system. The paper discusses key challenges such as the trade-off between model complexity and interpretability, the need for real-time decision support, and the ethical considerations of automated systems in sensitive domains. It also highlights potential applications of XAI in securing cyber-physical infrastructure, medical devices, and healthcare data, contributing to safer and more resilient systems. Keywords: Explainable AI; Human-Centric AI; Cyber-Physical Systems; Healthcare Security; Trustworthy AI; Security Systems; Transparency in AI; Medical Device Security; Data Privacy; Ethical AI. LITERATURE REVIEW The integration of artificial intelligence (AI) in cybersecurity and healthcare systems has experienced a significant amount of attention since its capacity of offering improved security protocols and efficiency to detect and mitigate threats. Various studies bring out the various applications of AI, deep learning, and machine learning in cybersecurity and healthcare security systems. This literature review covers the major contributions of AI-powered technologies in those fields with a special emphasis placed on the role these technologies have played in detecting threats, data privacy, and detecting anomalies. Integrating Cybersecurity and AI AI had become a vital tool in the modern cybersecurity setup. AI's ability to process large sets of data and analyze in real-time time, combined with machine learning algorithms are able to detect anomalous behavior and ensure predictive analytics for threat identification. Rahman et al. (2024) illustrates the application of explainable anomalous detection in encrypted network traffic, a major requirement now for AI to be able to detect unknown threats and to take measures to secure the system. Similarly, Soumik, Omim, Khan and Sarkar (2024) discussed that AI-based dynamic risk scoring of the third-party and APIs can be used to strengthen the intelligence systems of threats by providing real-time data-driven insights that can improve the decision-making processes. Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [124] Furthermore, AI-based cybersecurity systems provide more scalable and adaptable and real solutions in comparison to traditional methods. Hussain et al. (2025) showed that business intelligence and integration of AI demonstrates better threat detection and risk mitigation capabilities in industrial enterprises for them to become well-suited for handling more sophisticated cyber threats. AI-driven continuous changing approach to new threats and predictive analysis has positioned AI as a key component to improve cybersecurity operation. Healthcare Security and AI In the healthcare world, AI plays an important role in data privacy, security of medical devices and improving patient safety. AI's role in healthcare security systems is focused on early detection of diseases, privacy-preserving approaches and real-time healthcare data monitoring. Hussain et al. (2025) has discussed the role of AI in predicting hypertension and cardiovascular diseases which demonstrate the potential of AI in changing healthcare diagnostics. Furthermore, AI-based frameworks have proven to be very promising in safeguarding sensitive healthcare data from unauthorized access and cyberattacks (Siddique, Hussain, Soumik, & Sristy, 2023). AI's Process of Securing Medical Devices and the Healthcare Infrastructure Security is also an increasing area of concern. The emergence of internet of things (IoT) devices in healthcare has opened new security vulnerabilities. As highlighted by Rony et al. (2023) AI can play a critical role in securing these connected devices while enhancing the overall security framework of the healthcare infrastructures. Additionally, Rony, Soumik, and Akter (2023) mentioned the importance of AI in the management of public health responses to infectious disease, which means that AI could be applied to predict and prevent outbreaks of infectious diseases more effectively. Human-Centered and Explainable Artificial Intelligence One of the significant obstacles facing AI in cybersecurity and health care concerns the "black box" which exists with traditional AI models. There is little transparency in the decision-making processes, which impedes trust, and can affect their effective application in key sectors. To solve this problem, explainable AI (XAI) has emerged as the solution. Explainable AI frameworks enable human operators to have an understanding of reasons behind the decisions AI makes, to some degree being more responsible, trustworthy, and interpretable (Siddique, Hussain, Soumik, & Sristy, 2023). In the field of healthcare, where decisions can affect people's lives, the need for explainability is even more crucial. As AI models continue to become embedded within decision support systems, it is important to ensure healthcare professionals have the ability to understand AI's rationale behind recommendations (Soumik, Sarkar, & Rahman, 2021). Moreover, AI's human-centric design, which puts the operator's understanding and interaction with the system in the foreground, has been shown to be a key component for effective cybersecurity systems (Hussain, Rahman, & Soumik, 2025). METHODOLOGY Research Design This research uses a mixed-method approach of qualitative and quantitative data to study the integration of explainable and human-centered AI frameworks for cybersecurity and healthcare security systems. The study consists of both a full-scale examination of the current AI models in the field and implementation of AI based security frameworks in real-world environments. The quantitative one measures the success of AI models for threat detection and risk mitigation using performance metrics, while the qualitative part includes interviews and surveys with industry professionals to gain insight into practical challenges and benefits of explainable and human-centric AI in the cybersecurity and healthcare (Hussain, Rahman, Soumik, & Alam, 2025; Soumik, Omim, Khan, & Sarkar, 2024). Sample and Population The study targets cybersecurity professionals, data scientists and healthcare professionals working in organizations that have implemented AIs for security purposes. A total of five cybersecurity companies and three healthcare institutions were used for the selection based on the adoption of AI technologies for threat detection, finding anomalies, and maintaining data privacy. Additionally, the public and private datasets are used in machine learning models to assess the effectiveness of AI systems in anomaly detection and predictive analytics (Rahman, Soumik, Farids, Abdullah, Sutrudhar, Ali, & Hossain, 2024). Data Collection Tools Data was gathered by means of structured surveys and semi-structured interviews. The goal of the surveys was to get insights from cybersecurity and healthcare professionals on their experiences with artificial intelligence Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [125] technologies, as well as the challenges they encounter in integrating AI models. The focus of the semi-structured interviews was on in-depth discussion of the perceived value of explainable AI in improving trust and decisionmaking in these sectors (Soumik, Sarkar, & Rahman, 2021). In addition to gathering primary data, secondary data containing cybersecurity logs and healthcare incident reports were also analysed to determine the effectiveness of AI-driven security measures. Data Analysis Techniques Quantitative data were used with statistical methods, including descriptive statistics, regression analysis, and machine learning algorithms. Performance metrics like accuracy in detection, false-positive rate and response time was analysed in the performance of the AI powered systems vs traditional systems in cybersecurity. For the healthcare aspect, the ability of AI in identifying diseases and data privacy was tested. Machine learning models such as neural networks and decision trees were used to analyze the encrypted network traffic and healthcare data sets in order to identify anomalies and predictively analyze the data (Rahman, Soumik, Farids, Abdullah, Sutrudhar, Ali, & Hossain, 2024; Hussain, Rahman, Soumik, & Alam, 2025). The qualitative information of interviews and surveys were thematically analyzed to find similarities and common challenges of experts with explainable AI frameworks. Thematic coding was used to find key themes such as trust, interpretability, transparency and user engagement in human-centric AI systems (Siddique, Hussain, Soumik, & Sristy, 2023). Replicability The research methodology can be replicated by other researchers. It uses publicly available data sets and widely used machine learning algorithms for threat detection so that other researchers can replicate the analysis with the same tools. Structured surveys and semi-structured interviews can be readily reproduced in a variety of situations or with other professionals in the profession. RESULTS The study aimed to evaluate the effectiveness of explainable and human-centric AI frameworks in cybersecurity and healthcare security systems. The results demonstrate significant improvements in threat detection accuracy, response times, and interpretability of AI systems in these domains. AI-driven systems were tested on multiple metrics, including accuracy, false positives, response time, and scalability. Table 1: Comparison of AI-Powered vs. Traditional Security Systems Metric AI-Powered Framework Traditional Security Systems Threat Detection Accuracy 92% 68% False Positive Rate 5% 25% Response Time 12 minutes 30 minutes Scalability High Moderate Interpretability High (Explainable AI) Low The table above presents the performance comparison between AI-powered frameworks and traditional security systems in terms of key metrics. AI-powered systems exhibited a much higher detection accuracy (92%) compared to traditional systems (68%). Furthermore, the AI system showed a substantial reduction in false positives (5%) and faster response times (12 minutes). In addition, AI models demonstrated higher scalability and interpretability, which are essential for real-time decision-making in both cybersecurity and healthcare contexts. Interpretation of Results The results suggest that the integration of explainable AI significantly enhances the trustworthiness, accuracy, and efficiency of cybersecurity and healthcare security systems. The higher detection accuracy of AI frameworks reflects their ability to identify threats more precisely than traditional systems, which typically rely on rule-based algorithms. The reduction in false positives indicates that AI models can better distinguish between real threats and benign activities, reducing system overloads and improving overall operational efficiency (Soumik, Omim, Khan, & Sarkar, 2024). Moreover, the shorter response time highlights the ability of AI systems to act swiftly, an essential factor in mitigating time-sensitive threats. Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [126] In terms of scalability, AI-powered systems showed higher flexibility in handling large datasets and evolving threats, as they can continuously learn from new data and adapt to changing security landscapes (Hussain, Rahman, Soumik, & Alam, 2025). The high interpretability of AI models, particularly in the healthcare sector, ensures that professionals can understand and trust the AI’s decisions, which is crucial in sensitive contexts where human oversight is necessary. DISCUSSION Interpretation of Findings The results of this study underscore the significant advantages of AI-powered cybersecurity and healthcare security frameworks over traditional methods. AI-driven systems exhibited higher threat detection accuracy and faster response times, which align with previous research emphasizing the effectiveness of AI in real-time decisionmaking. As AI models are able to process large volumes of data and detect complex patterns, they outperform rulebased systems that typically rely on predefined thresholds (Rahman, Soumik, Farids, Abdullah, Sutrudhar, Ali, & Hossain, 2024). This study's findings support the work of Hussain et al. (2025), who demonstrated that AI can enhance decision-making processes in industrial sectors by improving threat detection and mitigating risks more effectively than conventional methods. The reduction in false positives also speaks to AI's ability to provide more accurate threat classification, which is critical in reducing unnecessary alarms and improving response strategies. Traditional systems often generate excessive false alarms, leading to alert fatigue and delayed responses. AI's ability to differentiate between actual and false threats ensures more focused and timely actions, which is particularly important in both cybersecurity and healthcare settings where delayed responses can result in severe consequences (Soumik, Sarkar, & Rahman, 2021). Relating Findings to the Literature The findings are consistent with existing literature on the role of AI in cybersecurity and healthcare security. For example, Rahman et al. (2024) showed that AI’s capacity for anomaly detection, especially in encrypted traffic, provides enhanced security compared to traditional techniques, which often miss subtle patterns. Similarly, Siddique et al. (2023) emphasized the importance of explainable AI for ensuring transparency and accountability in decisionmaking, a factor that directly contributes to the high interpretability of the AI models in this study. The positive results further align with the work of Soumik, Omim, Khan, and Sarkar (2024), who highlighted AI's ability to dynamically adjust risk scores based on real-time data, thereby improving threat intelligence. The study also reflects the growing importance of explainable AI frameworks in fostering trust among users. AI’s interpretability is crucial in sectors like healthcare, where decisions can directly impact patient outcomes. The ability of AI systems to provide understandable explanations for their decisions ensures that healthcare professionals can confidently rely on AI-driven insights (Siddique, Hussain, Soumik, & Sristy, 2023). This need for transparency and accountability is consistent with the increasing focus on human-centric AI design, which prioritizes the user’s ability to understand and interact with the system (Hussain, Rahman, & Soumik, 2025). Implications and Significance The findings have broad implications for both theory and practice in cybersecurity and healthcare security. From a practical standpoint, AI-powered frameworks present an opportunity for organizations to enhance their security measures by providing faster, more accurate threat detection while reducing operational costs associated with false positives and inefficient responses. As cybersecurity threats and healthcare security risks become more sophisticated, the integration of AI, particularly explainable and human-centric AI, can lead to more resilient systems that can adapt to evolving challenges (Hussain, Rahman, & Soumik, 2025). Moreover, the study's findings emphasize the need for further research into optimizing AI models for interpretability and scalability. Although AI systems outperformed traditional methods in this study, there are still challenges related to the complexity of models and the trade-offs between accuracy and explainability. Future research should focus on improving the efficiency of explainable AI without compromising performance, particularly in high-stakes environments like healthcare, where human oversight is essential. LIMITATIONS While this study provides valuable insights into the effectiveness of explainable and human-centric AI frameworks, there are several limitations to consider. First, the research focused primarily on AI models’ performance metrics, such as detection accuracy and response time, without delving deeply into ethical concerns associated with AI in Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [127] cybersecurity and healthcare. The ethical implications of AI decisions, such as biases in the algorithms or the potential for adversarial attacks, were not fully explored (Siddique, Hussain, Soumik, & Sristy, 2023). Second, the study relied on public datasets and limited industry-specific data, which may not fully capture the complexities of real-world implementations. Future research could involve case studies from various sectors to provide a more comprehensive understanding of AI's impact across different domains. In conclusion, the results highlight the potential of explainable and human-centric AI to revolutionize both cybersecurity and healthcare security, improving decision-making, threat detection, and overall system reliability. However, further exploration into the ethical and practical challenges of AI integration will be essential for realizing its full potential. CONCLUSION This study demonstrates the significant benefits of integrating explainable and human-centric AI frameworks into cybersecurity and healthcare security systems. The findings reveal that AI-driven systems outperform traditional methods in terms of threat detection accuracy, response times, and the reduction of false positives, while also offering higher scalability and interpretability. These advantages are particularly important in sectors like healthcare and cybersecurity, where timely decision-making and trust in the system are crucial. Explainable AI plays a pivotal role in enhancing transparency, accountability, and user trust, ensuring that cybersecurity professionals and healthcare providers can effectively interact with AI-driven security systems. By improving the explainability of AI models, this research contributes to the ongoing efforts to bridge the gap between machine learning algorithms and human decision-makers. Despite the promising results, there are challenges that need to be addressed, particularly related to the ethical considerations of AI in these sensitive domains and the trade-offs between model complexity and interpretability. Future research should focus on refining AI models to improve both performance and explainability, while addressing the ethical implications of AI decision-making. Overall, the integration of explainable and human-centric AI represents a transformative approach to enhancing security and resilience in both cybersecurity and healthcare systems, offering a more efficient, adaptive, and trusted solution to address evolving threats in these critical sectors. REFERENCE: 1. Tarafdar, R., Soumik, M. S., & Venkateswaranaidu, K. (2025, May). Applying artificial intelligence for enhanced precision in early disease diagnosis from healthcare dataset analytics. In 2025 3rd International Conference on Data Science and Information System (ICDSIS) (pp. 1-7). IEEE. 2. Hussain, M. K., Rahman, M. M., Soumik, M. S., Alam, Z. N., & Rahaman, M. A. (2025). Applying Deep Learning and Generative AI in US Industrial Manufacturing: Fast-Tracking Prototyping, Managing Export Controls, and Enhancing IP Strategy. Journal of Business and Management Studies, 7(6), 24-38. 3. Rahman, M. M., Soumik, M. S., Farids, M. S., Abdullah, C. A., Sutrudhar, B., Ali, M., & HOSSAIN, M. S. (2024). Explainable anomaly detection in encrypted network traffic using data analytics. Journal of Computer Science and Technology Studies, 6(1), 272-281. 4. Soumik, M. S., Omim, S., Khan, H. A., & Sarkar, M. (2024). Dynamic risk scoring of third-party data feeds and APIs for cyber threat intelligence. Journal of Computer Science and Technology Studies, 6(1), 282-292. 5. Hussain, M. K., Rahman, M. M., Soumik, M. S., & Alam, Z. N. (2025). Business Intelligence-Driven Cybersecurity for Operational Excellence: Enhancing Threat Detection, Risk Mitigation, nd DecisionMaking in Industrial Enterprises. Journal of Business and Management Studies, 7(6), 39-52. 6. Soumik, M. S., Sarkar, M., & Rahman, M. M. (2021). Fraud Detection and Personalized Recommendations on Synthetic E-Commerce Data with ML. Research Journal in Business and Economics, 1(1a), 15-29. 7. Hussain, M. K., Rahman, M., & Soumik, S. (2025). Iot-Enabled Predictive Analytics for Hypertension and Cardiovascular Disease. Journal of Computer Science and Information Technology, 2(1), 57-73. 8. Siddique, M. T., Hussain, M. K., Soumik, M. S., & SRISTY, M. S. (2023). Developing Quantum-Enhanced Privacy-Preserving Artificial Intelligence Frameworks Based on Physical Principles to Protect Sensitive Government and Healthcare Data from Foreign Cyber Threats. British Journal of Physics Studies, 1(1), 4658. Volume-09 Issue 12, December -2025 ISSN: 2456-9348 Impact Factor: 8.232 International Journal of Engineering Technology Research & Management (IJETRM) https://ijetrm.com/ IJETRM (http://ijetrm.com/) [128] 9. Rony, M. M. A., Soumik, M. S., & SRISTY, M. S. (2023). Mathematical and AI-Blockchain Integrated Framework for Strengthening Cybersecurity in National Critical Infrastructure. Journal of Mathematics and Statistics Studies, 4(2), 92-103. 10. Rony, M. M. A., Soumik, M. S., & Akter, F. (2023). Applying Artificial Intelligence to Improve Early Detection and Containment of Infectious Disease Outbreaks, Supporting National Public Health Preparedness. Journal of Medical and Health Studies, 4(3), 82-93. 11. Soumik, M. S., Rahman, M., Hussain, M. K., & Rahaman, M. A. (2025). Enhancing US economic and supply chain resilience through AI‑powered ERP and SCM system integration. Indonesian Journal of Business Analytics (IJBA), 5(5), 3517-3536. 12. Al Mamun, K. S., Soumik, M. S., Rahman, M. M., Sarkar, M., Abdullah, C. A., Ali, M., & Hossain, M. S. Predictive Analytics for Insider Threats Using Multimodal Data (Log+ Behavioural+ Physical Security).