AI-Augmented Safety Management in Construction

International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 DOI: 10.5121/ijcsit.2025.17507 107 AI-AUGMENTED SAFETY MANAGEMENT IN CONSTRUCTION Abdul Faisal Mohammed 1,Shahnawaz Mohammed 1, Abdul Raheman Mohammed 2, Syed Abdullah Kamran 1 1 Department of Information Technology, Trine University, MI, USA 2 Department of Information Technology, Lindsey Willson College, KY, USA ABSTRACT The construction sector continues to be one of the world's most hazardous, with high rates of accidents fuelled by multicomponent site dynamics, extensive use of heavy equipment, and unstable human behaviour. Conventional safety management methods, although essential, are generally reactive and fall short in offering real-time hazard perception or forecasting risk assessment. Recent advancements in Artificial Intelligence (AI) provide revolutionary opportunities to enhance safety performance through anticipatory, automated, and evidence-based decision-making. This article explains how AI techniques—ranging from computer vision for PPE detection and unsafe behaviour recognition, to wearable sensor analysis for fatigue and stress monitoring, to predictive machine learning models for incident prediction—can significantly enhance construction safety management. Furthermore, the combination of AI with Building Information Modelling (BIM) and digital twin technology allows for real-time hazard mapping, safety scenarios through simulation, and end-to-end synchronization between the virtual and physical worlds. This paper proposes a complete AI-based safety paradigm that harmonizes multimodal data sources, edge analytics, and interpretable predictive models to close the risk mitigation gap with worker privacy and trust. Data quality anomalies, model generalization, alert fatigue, and surveillance implications in terms of ethics are also addressed with responsible deployment practices. AI will eventually be able to shift construction safety from reactive compliance to preventive intervention, reducing incidents and safer conditions. KEYWORDS Artificial Intelligence (AI), Construction Safety, Safety Management, Computer Vision, Wearable Sensors, Predictive Analytics, Digital Twin, Building Information Modelling (BIM), Personal Protective Equipment (PPE), Hazard Prediction, Fatigue Monitoring, Edge Computing, Multimodal Data Fusion, Explainable AI, Risk Mitigation. 1. INTRODUCTION Construction is one of the riskiest occupations in the world, with a high proportion of occupational fatalities and injuries yearly. The employers are subjected to a range of risks such as fall from height, electric shock, struck by plant, and exposure to toxic chemicals. The dynamic process flow, fluidic organization on-site, and the number of contractors on-site, which demonstrate the construction process dynamic nature, induce risk control [1]. Traditional on-site safety control by surveillance, comprehension, and written documentation is key to risk avoidance but needs nominally reactive forms. They are human-observation-based systems and prone to subjectivity, latency, and situational awareness shortcomings. Artificial Intelligence (AI) is a new-age technology that has developed significantly in recent years and has immense potential to reshape construction safety management [2]. AI technologies, International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 18 aided by computer vision, machine learning, natural language processing, and sensor analytics, have the potential to provide real-time threat detection, risk calculation using predicted inputs, and proactive intervention approaches. For instance, adherence to PPE or dangerous postures is detectable by computer vision software without mechanical usage of observation cameras, while fatigue and stress levels leading to human errors are measurable by equipment sensors fixed on equipment or wearable sensors [3]. Similarly, analysis of past safety history, weather, and labor information as predictive analysis can forecast probability of accident such that it becomes simpler for the supervisors to correct the errors in advance. Other than individual uses, AI with BIM and digital twins offers a holistic view in the guise of digital replication of the building site in real time [4]. Integration offers dynamic visualization of risk, simulation of scenarios, and enhanced coordination among stakeholders. AI-driven safety systems not just enhance monitoring for compliance but also facilitate data-driven decisionmaking and thus change the paradigm from response to compliance to prevention [5].While such optimistic progress has indeed been achieved, there remain issues to be addressed. Issues of data privacy, system dependability, compatibility, and worker acceptance need to be overcome prior to large-scale implementation. In addition, explainable AI systems are essential to build confidence among site managers, workers, and regulators [6]. With its unique assessment of multimodal AI deployments—computer vision, sensor analytics, predictive modeling, and BIM/digital twin frameworks—against operational and ethical constraints, this survey provides the first comprehensive comparative analysis of AI-based safety management techniques in construction from both technical and organizational perspectives. Unlike prior evaluations that focused just on computer vision, this survey synthesizes multi-layer architectures, identifies specific implementation obstacles, and highlights emerging real-world concerns (privacy, scalability, and trust). This enables researchers and practitioners to work toward safer, more scalable, and morally responsible solutions. The application of AI-augmented safety management in the construction sector is examined in this survey using peer-reviewed research, industry case studies, and significant technological advancements from 2019 to 2025. Wearable sensor analytics, edge computing, computer vision for PPE detection, predictive risk modeling (deep neural networks, gradient boosting, and random forests), and integration with BIM and digital twin platforms are among the AI technologies covered in the study. Methods such as strictly conventional statistical forecasting, non-AI sensor deployments, and rule-based expert systems are not included. North America, Europe, and a few studies from Asia-Pacific are included in the geographical coverage, with the exclusion of low-resource environments where AI adoption has not yet started. Only publications written in English and supported by strong experimental or deployment data will be taken into consideration. 2. RELATED WORK A comprehensive search was conducted using databases like Web of Science, IEEE Xplore, and Scopus to locate reviewed literature. Terms such as "predictive safety analytics," "wearable fatigue monitoring," "AI construction safety," "computer vision PPE detection," and "BIM digital twin safety" have been used. Papers must offer verified technology frameworks, comparative deployment analyses, or original experimental findings in order to satisfy the inclusion requirements. Technology forecasts, opinion pieces, and case studies with no observable outcomes were not included. Before the full-text examination started, duplicates were eliminated. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 19 2.1. Computer Vision for PPE and Behaviour Detection Computer vision is becoming increasingly one of the most researched AI fields in construction safety. Models like YOLO, Faster R-CNN, and Efficient Det are trained on images from sites to identify PPE compliance (helmets, goggles, vests) and dangerous activities like wrong lifting or working in close proximity to restricted areas [7]. Models work best in controlled conditions but malfunction in low light, heavy occlusion, and uncontrolled camera orientations. Autonomous monitoring of PPE, however, eliminates considerable dependency on visual inspection and offers instantaneous monitoring of large-scale sites [8]. 2.2. Wearable and Physiological Monitoring Wearable devices with internal inertial measurement units (IMUs), heart rate, and electrodermal activity sensors are increasingly utilized for capturing worker fatigue, stress, and posture stability [9]. Machine learning algorithms from wearable data have demonstrated promising potential for pre-incident detection of hazardous states, e.g., drowsiness or overwork. Incorporation of wearables into regular working practices has been successful in pilots but worker acceptance, device wearability, and data confidentiality continue to be significant issues. 2.3. Predictive Analytics and Risk Prediction Predictive safety analysis blends incident history data, environmental factors, and project information to predict likely hazards [10]. Gradient boosting, random forests, and deep neural networks are just a few of the models that can detect indicators of serious injuries and fatalities (SIFs) [11]. Such techniques enable supervisors to plan in advance to deploy resources, thus channelling training, monitoring, or interventions to high-risk crews and tasks. Unavailability of large-scale, labelled data, however, limits the applicability of such models. 2.4. BIM and Digital Twin Integration AI integration with Building Information Modelling (BIM) and digital twins enables dynamic synchronisation of virtual and physical sites [12]. It is feasible to model dangerous scenarios, track proximity of workers from hazard areas, and schedule work in safer work sequences using safety analytics cast over 3D models. Digital twin–based systems enable continuous openness and facilitate early intervention as well as collaborative decision-making. While extremely promising, such systems are high in terms of computational power requirements and high-level hardware and software platform interoperability [13]. 2.5. Industrial Deployments and Case Studies Practical applications in the real world confirm that safety management with AI is no longer in the experimental phase [14]. Some of the big contractors use AI-enabled safety platforms to monitor compliance, predict risks, and minimize near misses. Case studies indicate a reduction in safety breaches and enhanced risk communication. Industry uptake is yet to move beyond the infancy stage, though, with impediments such as cost, integration complexity, and reluctance from trade unions due to fear of monitoring [15]. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 20 Table 1: Overview of Related Work in AI-Supported Construction Safety 3. SUGGESTED INTEGRATED FRAMEWORK With a view to harnessing the potential of AI for ensuring construction safety, this paper suggests an AI-Enriched Safety Management Framework that combines sensing technologies, real-time analytics, predictive modelling, and decision support [16]. The suggested framework is intended to be modular, scalable, and privacy-aware so it can be universally applied to various construction settings. The comparison table below summarizes major AI approaches: Table 2: Comparison of major AI approaches International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 21 3.1. Sensing and Data Acquisition Layer A multimodal sensing environment is the core of the framework. Fixed, portable, and dronebased video cameras record activities at the site for computer vision–driven PPE and behaviour tracking. Physiological and motion signals in real-time are provided by wearable sensors, such as heart rate, accelerometers, and IMUs, for fatigue and stress analysis. Concurrently, IoT-based environmental sensors track gas, noise, dust, and temperature [17]. These heterogeneous streams of information are augmented by BIM models, crew rosters, and a record of safety history, which provide a wealth of data context for safety analytics. 3.2. Edge Processing and Local Analytics Layer Due to the enormous amount of sensor data, edge computing hardware installed on-site conducts in situ preprocessing [18]. For example, video streams are processed by small deep models (e.g., YOLOv8) to detect PPE infringements, and wearable sensors compute fatigue indices from raw measurements. Edge analytics minimize latency, save bandwidth, and maintain confidentiality through anonymized event data transmission to the cloud. 3.3. Fusion and Predictive Modelling Layer At the hybrid or cloud level, multimodal data are fused by the system via techniques like crossmodal Transformers and ensemble learning [19]. Predictive models analyse context and temporal trends to predict incident risk levels at site, employee, and crew level. Predictive models provide risk heatmaps and accident probability scores for accidents in target zones or tasks. To enable adoption, the design incorporates explainable AI (XAI) features like SHAP values or attention visualizations to enable managers to see for what reasons predictions are being generated for the risks [20]. 3.4. Decision Support and Intervention Layer Data generated is returned to stakeholders in an interactive dashboard and BIM-integrated visualizations [21]. Supervisors can see real-time alerts, monitor patterns of compliance, and test "what-if" safety interventions in the digital twin environment. Employees are provided with audio or haptic notifications on wearables when they enter at-risk zones, and equipment shutdowns can automatically be initiated for imminent danger. Perhaps most importantly, the system includes a team's feedback loop, where outcomes of interventions are monitored repeatedly to retrain and refit prediction models. AI-Augmented Safety Management Framework Proposed [22]. • Layer 1 (Sensing): Cameras, drones, wearables, IoT sensors, BIM data sources. • Layer 2 (Edge Analytics): PPE detection, fatigue rating, environmental preprocessing on site devices. • Layer 3 (Fusion & Prediction): Cloud/hybrid platform with multimodal fusion, predictive algorithms, explainable AI. • Layer 4 (Decision Support): Dashboards, BIM/Digital Twin overlays, worker notifications, automated interventions. • Feedback Loop: Continuous learning from outcomes. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 22 Figure 1. Proposed AI-Enhanced Safety Management Framework 4. METHODOLOGY Organizational implementation of applying AI-aided safety management on building construction sites goes through four steps: data collection and preprocessing, model building, system integration, and testing [23]. Technical demands as well as concrete difficulties for sound and ethical implementation are explained in each step. 4.1. Data Preprocessing and Gathering The foundation of the approach is based on strong and diversified data collection [24]. Information on construction sites are gathered from three primary sources • Visual data captured by cameras and drones (compliance with PPE, recognition of unsafe posture). • Physiological data from wearables (heart rate, accelerometers, EMG). • Contextual data like weather, equipment usage, and project schedule. Data preprocessing involves noise removal, anonymization (blurring of faces), and feature extraction. The image data are labelled with bounding boxes and key points, while the signals are transformed into fatigue indexes for the wearables. Oversampling and augmentation are applied to resample the data to counter class imbalance in the rare event cases [25]. 4.2. Model Development Artificial intelligence models are created to execute detection and prediction functions: • Computer Vision Models: YOLOv8 and Faster R-CNN to detect PPE and risky behaviour. • Wearable Analytics: Random Forest and LSTM models for predicting fatigue and stress classification. • Predictive Risk Models: Gradient boosting and deep neural networks to predict incidents with multimodal inputs. Explainability of the models is guaranteed through the use of SHAP values and attention maps to provide transparent decision-making [26]. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 23 4.3. System Integration Models run in a hybrid edge–cloud configuration [26]. Edge appliances perform local highfrequency processing (e.g., PPE checking), with the cloud reserved for more intensive predictive workloads (e.g., 24-hour incident prediction). Integration is facilitated by an integration engine that aggregates multimodal outputs to produce risk scores at worker, task, and site levels [27]. Outputs are inputted to dashboards and BIM-integrated visualizations such that supervisors can monitor risks in real time. 4.4. Evaluation and Validation It is assessed with technical performance metrics and field pilots: • Detection Models: Mean Average Precision (MAP), precision, recall, F1 score. • Predictive Models: Area Under Curve (AUC), accuracy, precision–recall balance. • Operational Metrics: Time-to-detection, false alarm rates, and incident reduction per 1,000 worker hours. Field pilots consist of controlled experiments (simulated PPE violations, fatigue conditions) and live deployment on active construction sites [28]. A linear process with four integrated phases: 1. Data Collection & Preprocessing → 2. Model Building (CV, Wearables, Prediction) → 3. Integration with System (Edge + Cloud, Fusion Engine, BIM Dashboard) → 4. Evaluation & Verification (Metrics + Field Trials). The process above shows the end-to-end pipeline from raw data to effective safety interventions. Figure 2. Methodology Line Diagram 5. DATASETS & BENCHMARKING The performance of AI-based safety management in construction relies heavily on diversity, representativeness, and quality of data sets [29]. Systematic benchmarking allows for reproducible and objective evaluation of AI models for generalizable outcome on a range of different project environments. 5.1. Visual Data-Based Computer Vision Tasks • Visual data forms the basis for AI-enabled hazard detection and PPE conformity checking. Sources are:Public Datasets: Public datasets like PPE Dataset, Safety Helmet Detection Dataset, and COCO construction subsets offer object detection through labelled images [30]. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 24 • Drone Footage: Construction site-specific datasets recorded on operational construction sites offer aerial views for hazard localization. • On-Site CCTV & Camera Streams: These datasets support recognition of unsafe behaviour in real-time and crowd movement monitoring [31]. In return for providing robustness, images are labelled with PPE equipment, dangerous behaviours (e.g., ladder misuse), and off-limit areas by bounding boxes and segmentation masks [32]. 5.2. Wearable Sensor Data Sets Wearable sensors offer physiological and movement data needed to track worker fatigue and stress. • Heart Rate & ECG Datasets: For detection of stress and foretelling initial fatigue. • Accelerometer & Gyroscope Data: Receive real-time worker posture, slips, or falls. • Environmental Sensors: Measure exposure to heat, noise, and vibration. These observations are compared to prevailing health standards to find deviations that are associated with hazardous situations [33]. 5.3. Multimodal & Synthetic Datasets Combining multimodal datasets (visual + wearable + contextual) provides an integrated view of the risks to safety [34]. Synthetic datasets from simulation within digital twins or simulated environments also serve a helpful purpose in compensating for the lack of unusual accident cases, i.e., crane crashes or high-rise building falls. 5.4. Benchmarking Protocols Benchmarking employs the shared evaluation measures: • Computer Vision Tasks: Mean Average Precision (MAP), Intersection over Union (IoU), and detection latency. • Wearable Analytics: Accuracy, F1 score, and confusion matrices for stress/fatigue prediction. • Multimodal Fusion Models: Area Under Curve (AUC), risk score calibration, and decrease in false alarms. Cross-validation across multiple datasets ensures generalization. Furthermore, citation of baseline models (rule-based safety monitoring, conventional monitoring) emphasizes the value added by AI [35]. 5.5. Ethical & Privacy Considerations Collection of the dataset and benchmarking gives origins of privacy, consent, and bias in data issues [36]. The identities of employees are anonymized by face blurring and encrypting data, and sampling of the dataset is done to give balanced representation based on gender, roles, and conditions to avoid biased prediction. Pie chart indicating dataset composition used in AI-augmented safety: International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 25 • Visual Data (40%) • Wearable Sensor Data (30%) • Contextual Data (15%) • Synthetic/Digital Twin Data (15%) This underlines the ratio of actual-world and simulated data utilized in underpinning AI benchmarking. Figure 3. Dataset Distribution Pie Diagram 6. CASE STUDY: HYPOTHETICAL DEPLOYMENT (ILLUSTRATIVE) To underpin the applicability of using the suggested AI-based safety management framework in actual-world practice, a hypothetical deployment is represented for a high-rise building construction site with multiple contractors, heavy plant, and complicated scheduling [37]. 6.1. Project Context The project consists of a 40-storey office building in a city with tight schedules and conflicting packages [38]. Hazard risks are height fall, crane collision, and fatigue-based accidents. About 300 workers report to work on site every day, consisting of skilled labour, subcontractors, and machinery operators. 6.2. System Deployment The model is deployed in three phases: • Phase 1 – Installation of Sensors: Cameras are installed above scaffolding, entry points, and equipment areas. Fatigue and posture tracking devices are made available to workers. IoT sensors monitor dust, noise, and temperature levels [39]. • Phase 2 – Integration with AI: Real-time PPE compliance is executed by edge devices, while multimodal data are analyzed using cloud models in an endeavor to create risk heatmaps. BIM overlie hazards are integrated on a digital twin in an effort to help visualize. • Phase 3 – Alerts & Feedback: Employees get haptic notifications on wearables for highrisk zones. Managers see predictive safety interventions and intervention suggestions on dashboards [40]. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 32 [2] Fitria, T. N. (2021, December). Artificial intelligence (AI) in education: Using AI tools for teaching and learning process. In Prosiding seminar nasional & call for paper STIE AAS (pp. 134-147). [3] Gattuso, A. N. (2021). Common issues of compliance with personal protective equipment for construction workers. [4] Azanaw, G. M. (2024). Revolutionizing Structural Engineering: A Review of Digital Twins, BIM, and AI Applications. Indian Journal of Structure Engineering (IJSE), 4(2), 1-8. [5] Usama, M., Ullah, U., Muhammad, Z., Islam, T., & saba Hashmi, S. (2024). AI-enabled risk assessment and safety management in construction. In Ethical Artificial Intelligence in Power Electronics (pp. 105-132). CRC Press. [6] Sharma, G., Karim, R., Samuelson, O., & Simu, K. (2023, June). A Conceptual Model for AIEnabled Digitalization of Construction Site Management Decision Making. In International Congress and Workshop on Industrial AI (pp. 145-159). Cham: Springer Nature Switzerland. [7] Prabha, M., Karuppasamy, M., Saraswathi, P., Poorani, K., & Aswin Kumar, S. (2024, November). Personal Protective Equipment Detector Using Computer Vision. In 2024 4th International Conference on Advancement in Electronics & Communication Engineering (AECE) (pp. 38-41). IEEE. [8] Gallo, G., Di Rienzo, F., Garzelli, F., Ducange, P., & Vallati, C. (2022). A smart system for personal protective equipment detection in industrial environments based on deep learning at the edge. IEEE Access, 10, 110862-110878. [9] Zhang, Y., Zheng, X. T., Zhang, X., Pan, J., & Thean, A. V. Y. (2024). Hybrid integration of wearable devices for physiological monitoring. Chemical Reviews, 124(18), 10386-10434. [10] Alotaibi, E. M. (2023). Risk assessment using predictive analytics. International Journal of Professional Business Review: Int. J. Prof. Bus. Rev., 8(5), 59. [11] Bayona, A., Bhandari, S., Hallowell, M., Sherratt, F., Bailey, J. M., & Upton, J. (2023). What is a serious injury? A model for defining serious injuries & fatalities. Professional Safety, 68(09), 22-30. [12] Nguyen, T. D., & Adhikari, S. (2023). The role of bim in integrating digital twin in building construction: A literature review. Sustainability, 15(13), 10462. [13] Bandi, A., Adapa, P. V. S. R., & Kuchi, Y. E. V. P. K. (2023). The power of generative ai: A review of requirements, models, input–output formats, evaluation metrics, and challenges. Future Internet, 15(8), 260. [14] Paleyes, A., Urma, R. G., & Lawrence, N. D. (2022). Challenges in deploying machine learning: a survey of case studies. ACM computing surveys, 55(6), 1-29. [15] Nissim, G., & Simon, T. (2021). The future of labor unions in the age of automation and at the dawn of AI. Technology in Society, 67, 101732. [16] Jarek, K. New Product Development Enriched with AI. In AI in Marketing (pp. 194-222). Routledge. [17] Sofi, A., Regita, J. J., Rane, B., & Lau, H. H. (2022). Structural health monitoring using wireless smart sensor network–An overview. Mechanical Systems and Signal Processing, 163, 108113. [18] Lujic, I., De Maio, V., Venugopal, S., & Brandic, I. (2021). Sea-leap: Self-adaptive and localityaware edge analytics placement. IEEE Transactions on Services Computing, 15(2), 602-613. [19] Ansari, M. F. Redefining Cybersecurity: Strategic Integration of Artificial Intelligence for Proactive Threat Defense and Ethical Resilience. [20] Mohammed, Z., Mohammed, N. U. M., Mohammed, A., Gunda, S. K. R., & Ansari, M. A. A. (2025). AI-Powered Energy Efficient and Sustainable Cloud Networking. Journal of Cognitive Computing and Cybernetic Innovations, 1(1), 31-36. [21] Alvarez, S. A., & Sachs, S. (2023). Where do stakeholders come from? Academy of Management Review, 48(2), 187-202. [22] Barua, C., Kabir, J. U. Z., Alam, K. R., Nabil, A. R., & Duman, S. (2025). AI-augmented agile project management in engineering: A framework for smart decision-making and risk mitigation. International Journal of Science and Research Archive, 15(3), 967-973. [23] Bhoj, K., Sharma, Y., Syed, A., Vasoya, A., & Thakur, R. (2023). Effective Construction Site Monitoring Using Artificial Intelligence (AI). In Smart Cities (pp. 116-131). CRC Press. [24] Padthe, Adithya, Rashmi Ashtagi, Sagar Mohite, Prajakta Gaikwad, Ranjeet Bidwe, and H. M. Naveen. "Harnessing federated learning for efficient analysis of large-scale healthcare image datasets in iot-enabled healthcare systems." International Journal of Intelligent Systems and Applications in Engineering 12, no. 10s (2024): 253-263. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 33 [25] Wu, L., Zhuang, J., Chen, W., Tang, Y., Hou, C., Li, C., ... & Luo, S. (2022). Data augmentation based on multiple oversampling fusion for medical image segmentation. PloS one, 17(10), e0274522. [26] Khadri, S. W., Mohammed, I. K., Rasheed, H., & Gunda, S. K. R. (2025). Adaptive Trade Exception Handling in Financial Institutions: A Reinforcement Learning Approach with Dynamic Policy Optimization. Journal of Cognitive Computing and Cybernetic Innovations, 1(1), 19-23. [27] Mulaydinov, F. M. (2024). Integrated information systems in industrial enterprises and business activities. University Research Base. [28] Musca, R., Vasile, A., & Zizzo, G. (2022). Grid-forming converters. A critical review of pilot projects and demonstrators. Renewable and Sustainable Energy Reviews, 165, 112551. [29] Mohammed, N. U., Mohammed, Z. A., Gunda, S. K. R., Mohammed, A., & Khaja, M. U. Networking with AI: Optimizing Network Planning, Management, and Security through the medium of Artificial Intelligence. [30] Khadri, W., Reddy, J. K., Mohammed, A., & Kiruthiga, T. (2024, July). The Smart Banking Automation for High Rated Financial Transactions using Deep Learning. In 2024 IEEE 3rd World Conference on Applied Intelligence and Computing (AIC) (pp. 686-692). IEEE [31] Chittoju, S. R., & Ansari, S. F. (2024). Blockchain’s Evolution in Financial Services: Enhancing Security, Transparency, and Operational Efficiency. International Journal of Advanced Research in Computer and Communication Engineering, 13(12), 1-5. [32] Dagli, R., & Shaikh, A. M. (2023). Cppe-5: Medical personal protective equipment dataset. SN Computer Science, 4(3), 263. [33] Begum, A., Mohammed, N., & Panda, B. B. (2024). Leveraging AI in health informatics for early diagnosis and disease monitoring. IARJSET, 11(12), 71–79. https://doi.org/10.17148/iarjset.2024.111205 [34] Zhang, D., Zhang, M., Zhang, H., Yang, L., & Lin, H. (2021, August). MultiMET: A multimodal dataset for metaphor understanding. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers) (pp. 3214-3225). [35] Chittoju, S. S. R., Kolla, S., Ahmed, M. A., & Mohammed, A. R. Synergistic Integration of Blockchain and Artificial Intelligence for Robust IoT and Critical Infrastructure Security. [36] Mohammed, A., Mohammed, N. U., Gunda, S. K. R., & Mohammed, Z. Fundamental Principles of Network Security. [37] Dickerman, B. A., Dahabreh, I. J., Cantos, K. V., Logan, R. W., Lodi, S., Rentsch, C. T., ... & Hernán, M. A. (2022). Predicting counterfactual risks under hypothetical treatment strategies: an application to HIV. European Journal of Epidemiology, 37(4), 367-376. [38] Padthe, Adithya, Ramya Thatikonda, and Rashmi Ashtagi. "Leveraging generative adversarial networks for cross-modal image processing." In Artificial Intelligence and Information Technologies, pp. 176-180. CRC Press, 2024. [39] Mohammed, A., Sultana, G., Aasimuddin, F. M., & Mohammed, S. (2025). Leveraging Natural Language Processing for Trade Exception Classification and Resolution in Capital Markets: A Comprehensive Study. Journal of Cognitive Computing and Cybernetic Innovations, 1(1), 14-18. [40] Saxena, V. (2024). Predictive analytics in occupational health and safety. arXiv preprint arXiv:2412.16038. [41] Isailovic, V., Peulic, A., Djapan, M., Savkovic, M., & Vukicevic, A. M. (2022). The compliance of head-mounted industrial PPE by using deep learning object detectors. Scientific Reports, 12(1), 16347. [42] Huber, J., Anzengruber-Tanase, B., Schobesberger, M., Haslgrübler, M., Fischer-Schwarz, R., & Ferscha, A. (2025). Evaluating User Safety Aspects of AI-Based Systems in Industrial Occupational Safety: A Critical Review of Research Literature. International Journal of Environmental Research and Public Health, 22(5), 705. [43] Erdener, B. C., Sergi, B., Guerra, O. J., Chueca, A. L., Pambour, K., Brancucci, C., & Hodge, B. M. (2023). A review of technical and regulatory limits for hydrogen blending in natural gas pipelines. International Journal of Hydrogen Energy, 48(14), 5595-5617. [44] Janamolla, Kavitha, Ghousia Sultana Sultana, Fnu Mohammed Aasimuddin, Abdul Faisal Mohammed, and Fnu Shaik Aqheel Pasha Pasha. "Integrating Blockchain and AI for Efficient Trade Exception Handling: A Case Study in Cross-Border Settlements." Journal of Cognitive Computing and Cybernetic Innovations 1, no. 1 (2025): 24-30. International Journal of Computer Science & Information Technology (IJCSIT) Vol 17, No 5, October 2025 34 [45] Mohammed, A. K., & Ansari, M. A. (2024). The Impact and Limitations of AI in Power BI: A Review . International Journal of Multidisciplinary Research and Publications (IJMRAP), , Pp. 2327, 2024. , 7(7), 24–27 [46] Syed, W. K., Mohammed, A., Reddy, J. K., & Dhanasekaran, S. (2024, July). Biometric Authentication Systems in Banking: A Technical Evaluation of Security Measures. In 2024 IEEE 3rd World Conference on Applied Intelligence and Computing (AIC) (pp. 1331-1336). IEEE. [47] Ahmed, N., Hossain, M. E., Rishad, S. S. I., Rimi, N. N., & Sarkar, M. I. (2021). Server less Architecture: Optimizing Application Scalability and Cost Efficiency in Cloud Computing. BULLET: Jurnal Multidisiplin Ilmu, 1(06), 1366-1380. [48] Orhan, M., & Ozdemir, T. Y. (2022). Development of Accountability Behavior Scale for School Administrators: A Validity and Reliability Study. International Online Journal of Educational Sciences, 14(5). [49] RAHEEM, MOHD ABDUL, and MOHAMMED AZMATH ANSARI. "INTELLIGENT AND TRUSTWORTHY 6G: AI-DRIVEN ARCHITECTURES, APPLICATIONS, AND SECURITY FRAMEWORKS." [50] Stahl, B. C. (2021). Ethical issues of AI. In Artificial Intelligence for a better future: An ecosystem perspective on the ethics of AI and emerging digital technologies (pp. 35-53). Cham: Springer International Publishing. [51] Sherefetdinova, E. (2024). Social issues in works of charles dickens. Talqin va tadqiqotlar ilmiyuslubiy jurnali, 2(56), 109-112. [52] da Silva Neto, R. M., Benjamim, C. J. R., de Medeiros Carvalho, P. M., & Neto, M. L. R. (2021). Psychological effects caused by the COVID-19 pandemic in health professionals: a systematic review with meta-analysis. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 104, 110062. [53] Chen, Y., Biswas, M. I., & Talukder, M. S. (2023). The role of artificial intelligence in effective business operations during COVID-19. International Journal of Emerging Markets, 18(12), 63686387. [54] Borges, A. F., Laurindo, F. J., Spínola, M. M., Gonçalves, R. F., & Mattos, C. A. (2021). The strategic use of artificial intelligence in the digital era: Systematic literature review and future research directions. International journal of information management, 57, 102225. [55] Su, L. C., Wu, X., Zhang, X., & Huang, X. (2021). Smart performance-based design for building fire safety: Prediction of smoke motion via AI. Journal of Building Engineering, 43, 102529. [56] Varshney, Y., & Ram, N. (2024). Weather Forecasting: A Systematic Review Using AI Approaches. Weather, 4(10). [57] Chinta, S. (2024). Edge AI for real-time decision making in IoT networks. International Journal of Innovative Research in Computer and Communication Engineering, 12(9), 11293-11309. [58] Ueno, T., Sawa, Y., Kim, Y., Urakami, J., Oura, H., & Seaborn, K. (2022, April). Trust in human-AI interaction: Scoping out models, measures, and methods. In CHI conference on human factors in computing systems extended abstracts (pp. 1-7). [59] Suhag, D. (2024). Regulatory and Ethical Considerations. In Handbook of Biomaterials for Medical Applications, Volume 2: Applications (pp. 355-372). Singapore: Springer Nature Singapore. [60] Eneyew, D. D., Capretz, M. A., & Bitsuamlak, G. T. (2022). Toward smart-building digital twins: BIM and IoT data integration. IEEE access, 10, 130487-130506. [61] Delardas, O., Kechagias, K. S., Pontikos, P. N., & Giannos, P. (2022). Socio-economic impacts and challenges of the coronavirus pandemic (COVID-19): an updated review. Sustainability, 14(15), 9699. [62] Rahrooh, A., Garlid, A. O., Bartlett, K., Coons, W., Petousis, P., Hsu, W., & Bui, A. A. (2024). Towards a framework for interoperability and reproducibility of predictive models. Journal of Biomedical Informatics, 149, 104551.