Smart Waste Management in the Age of Industry 4.0: IoT, AI and Blockchain Applications in Circular Economy Systems

 Corresponding author: Ankur Singh Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0. Towards Responsible AI in Autism Care: A Multi-Modal Federated-Edge Framework for Real-Time Behavioral Support Ankur Singh 1, * and Sajjadur Rahman 2 1 School of Computer Science, University of North America, Fairfax VA, USA. 2 School of Computing and Digital Technology, Birmingham City University, Birmingham B5 5JU, United Kingdom. World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 Publication history: Received on 21 August 2025; revised on 01 October 2025; accepted on 04 October 2025 Article DOI: https://doi.org/10.30574/wjarr.2025.28.1.3364 Abstract Autism spectrum disorder (ASD) is a condition that is mostly characterized by unpredictable behavioral outbursts that need to be constantly observed and addressed in time. The conventional AI-based clinical decision-support systems (CDSS) have enhanced precision medicine and behavioral health but often do not help to capture the multi-modal nature of the autism behaviors. In addition, the issues relating to data privacy, latency of the system and clinician trust limit their adoption. In this paper, the challenges have been discussed by providing a federated-edge AI framework that coalesces streams of speech, motion, and physiological data into an explainable CDSS. The system, in contrast to the centralized models, utilizes edge intelligence to provide low-latency processing, federated learning with differential privacy to provide secure collaboration, and explainability dashboards to build clinical trust. Assessments of synthetic data show improvements in accuracy, lower latency, and usability, and prove the potential of the model as a scalable, ethical, and transferable model of autism care, behavioral health, and precision medicine. Keywords: Autism spectrum disorder; Multi-modal AI; Edge intelligence; Federated privacy; Explainable clinical AI; Ethical deployment. 1. Introduction The problem of autism care has been relying on human observation and care expertise to predict and control behavioral escalation over decades. Although this method has proven to be successful in some situations, it is subjective by nature, imprecise and confined to the presence of trained caregivers. These limitations have led to the desire to develop technological solutions that can augment the expertise of humankind with the use of data-driven knowledge. In that aspect, the advent of AI-based clinical decision-support systems (CDSS) has proved to be very promising, providing additional precision medicine, mental health, and workforce planning opportunities [1-3]. Although these improvements have been made, AI has not been widely used in the management of autism. A significant weakness is that existing systems seldom have the ability to record the multi-modality of behavioral escalation, which in most cases is expressed as a variation of speech, abnormal pattern of movement, and alteration of a certain physiological condition, all of which happen at the same time. Treatment systems that isolate these signals do not acknowledge the complexity and interdependence of behaviors associated with autism making them less effective in practice. Meanwhile, centralized AI models are extremely problematic when it comes to responsible deployment. They often invade the privacy of patients by making them send and store sensitive health information on external servers. Their dependence on cloud-based computation is a source of a latency that compromises the real-time responsiveness World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 393 required over the behavioral crisis. Moreover, the absence of clear explanations of the AI-driven predictions diminishes the trust of clinicians in this matter thus complicating its adoption in the clinical setting where accountability and explainability are critical. These barriers have been proposed in recent studies with partial solutions. Differentiated privacy Federated learning is a method that allows training in institutions with the distribution of sensitive data without leaving the local environment [4]. Similar advances in monitoring IoT have indicated opportunities in the ability to capture and identify early behavioral changes using wearable sensors [8,9]. In addition to technical innovations, transparency artifacts and usability checklists have also underscored the importance of trust and interpretability to clinician adoption [7,13]. These are in line with larger perspectives in human-centered AI, which highlight accountability, fairness, and inclusivity as being required to apply AI with ethics [15]. Expanding on these strands, this paper suggests creating a multi-modal federated-edge AI architecture that will combine speech, motion, and physiological data into a safe, explicable, and real-time CDSS. The framework would eliminate the limitations of centralized models by integrating the privacy-sensitive federated learning, reduction latency based on edges, and explainability dashboards to clinicians. This way, it promotes a responsible and clinically pertinent autism care AI architecture, and more generally, behavioral health and precision medicine. 2. Related Work 2.1. Precision Medicine and quantum-AI Recent developments in fields of quantum-enhanced artificial intelligence have revolutionized the biomedical discovery by accelerating drugs development, protein folding, and personalised modeling therapies [1]. These discoveries will show that high-performance computational frameworks can go beyond molecular medicine to behavioral health, in which analogous methods could be used to create real-time personalized care in autism. 2.2. Fraud Analytics and Anomaly Detection Methods that were created to combat fraud have been reused in the medical field to identify anomalies in multidimensional streams of behavior [2]. With autism care, the techniques are especially useful in detecting minor, but vital escalation signs in instances where speech, movement, or bodily abnormalities occur and are indicative of behavioral crises. 2.3. IoT-Based Autism Monitoring The use of wearables and IoT devices in healthcare has increased at an accelerated pace, especially in the area of autism monitoring. Constant monitoring of the data gathered into physiological measurements and motion capturing allows to timely identify stress or any abnormal behavior [8,9]. It is on the basis of these technologies that the scalable noninvasive based monitoring systems can be developed that can support both the caregivers and the clinicians in particular. 2.4. AI in Crisis-Response Models Artificial intelligence has also been effective in crisis-management situations, where it is necessary to detect and intervene in time. It has been demonstrated how AI can be used in the prevention of overdose and the expansion of behavioral health services by becoming part of the national 988 hotline continuum [10,11]. It is through such work that the possibility of AI systems being used as front-line supports in high-stakes health situations can be identified. 2.5. Collaborative Learning and Cybersecurity Systemically, the idea of federalized learning models with disparate privacy has produced solid approaches to protecting delicate health data in training distributed model [4]. In addition to this, data-centric cybersecurity solutions address the risk of interconnected medical devices, securing that healthcare AI is safe even in more hyper-digitized environments [12]. 2.6. Organizational AI and Workforce Planning They also apply AI at organizational level, with workforce planning decision-support tools applied to increase efficiency and human resource allocation optimization in clinical settings [13]. The tools are especially applicable in the autism care where the quality of intervention depends directly on the availability of caregivers. World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 394 2.7. Ethical and Responsible AI At the international level, AI ethics frameworks pay significant attention to the need to be fair, transparent, accountable, and inclusive [14,15]. The values are not merely ideals but have to be put into practice to allow AI to be incorporated into healthcare responsibly, where trust is essential and no groups of people have unequal results. 2.8. Synthesis All these studies show a significant advance in precision AI, anomaly detection, IoT monitoring, crisis management, privacy-preserving learning, and ethical AI frameworks. However, there is an obvious gap: there are not many such endeavors that allow merging multi-modal behavioral information with federated privacy and explainability into a single CDSS that draws attention to autism. This is the main point of the current work as it aims to address this gap. 3. Methodology 3.1. Framework Design The suggested architecture (Figure 1) will be organized in the form of a three-layer federated-edge model that would enable clinical decision support (autism care) to be secure, real-time, and interpretable. 3.2. Edge IoT Monitoring Wearable and mobile sensors in the first layer record multi-modal data streams, such as speech signal, motion signal, and physiological signals, such as heart rate and galvanic skin response. Finally, keeping calculation at the edge means that data can be pre-processed on the edge to minimize noise and latency to detect anomalies almost in real-time and allows the system to rely less on centralized clouds. 3.3. Federated Privacy-Preserving Learning The second layer uses a federated learning framework that uses differential privacy protection [4]. Training of the local models is done in participating institutions and only encrypted parameters or gradients are sent to a central aggregator. This makes sure that data of sensitive patients does not leave its original source hence keeping confidentiality yet allowing collaborative intelligence. The protection against the possible data reconstruction or leakage is also enhanced by the use of secure aggregation protocols. 3.4. Explainability Dashboard The third layer gives clinicians the ability to view an explainable AI dashboard that encompasses feature-attribution methods (e.g., SHAP values, confidence scores) to increase the interpretability. The prediction is accompanied by context-related information, including whether the anomalies were provoked mainly by speech abnormalities, movement patterns or physiological signs of stress. Along with that, the dashboard will have a workforceknowledgeable module, which will provide recommendations regarding caregiver assignments, alerts about escalation, and crisis-response guidelines. The design will guarantee not only technically sound but also clinically practicable and reliable outputs. World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 395 Figure 1 Circular federated edge AI workflow for autism care 3.5. Data Sources As a test of the framework, three datasets were used: • Artificial Multi-Modal Autism Data. The data is simulated with the involvement of 150 participants (including speech, motion, and physiological signals). This data is representative of behavioral complexity in the real world and allows privacy. • Workforce Scheduling Data. A catalog of 300 entries that represent caregiver shifts and staffing levels that will be used to confirm the workforce planning aspect of the system. • Systems on a chip: anomaly Detection Benchmark Dataset. An irregular case dataset (200) based on fraud analytics approaches used to test the performance of anomaly detection in the task of predicting escalation. 3.6. Evaluation Metrics The framework was evaluated based on the mixture of technical and usability-based evaluations: • Predictive Performance. To assess the effectiveness of the model in identifying behavioral anomalies and escalation events, Accuracy, Recall and Precision were determined. • Latency. The edge-based predictions and cloud-only baselines were compared by measuring inference time (milliseconds), which is the measure of the benefits of local computation. • Trust and Usability on the part of clinicians. Clinicians and caregivers were surveyed in a survey-based assessment (Likert scale) to determine the interpretability of the system, its trustworthiness, and its ability to be integrated into clinical processes. World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 396 4. Results Table 1 The framework outperformed a cloud-only baseline 5. Discussion 5.1. Novelty The current paper presents a federated-edge multi-modal AI which proposes speech, motion and physiological signals to a single autism-oriented clinical decision-support system. The proposed framework integrates the three dimensions latency, privacy, and modality integration at the same time unlike earlier studies that focused either on one dimension such as autism behaviors monitored using IoT [8] or federated privacy-preserving architectures [4]. In this way, it fills a gap in the literature that is crucial since it provides a more holistic approach to autism care based on ethics. 5.2. Trust The black-box problem is considered to be one of the greatest obstacles to AI usage in healthcare since clinicians are not eager to make decisions based on the predictions that they cannot explain. To address this, the suggested system would include explainability dashboards whose features would involve the use of feature-attribution methods in explaining decision pathways [7]. Besides increasing interpretability, such dashboards combine workforce-sensitive suggestions, so that predictions are accompanied by practical clinical insights [13]. By adhering to the principles of human-centered AI, such as transparency, accountability, and fairness, the framework contributes to further the design philosophy in which clinician trust is in the heart of the framework [15]. 5.3. Clinical Relevance The system has good chances of real time escalation management in autism treatment whereby real time detection and intervention are very critical. The framework will minimize false negatives by incorporating several modalities of behavior to enhance the accuracy of early detection. Furthermore, its design is complementary to the current crisisresponse systems like the 988 hotline continuum [10], highlighting its ability to expand beyond autism care to the wider use of behavioral health, workforce planning, and precision medicine [3]. This flexibility makes the framework a flexible solution that can be used to handle various healthcare issues. 5.4. Limitations Although the findings are encouraging, it is necessary to note a number of constraints. It is limited by the use of synthetic datasets, which are not able to fully reflect the complexity of the real-world clinical environment. The solution to this is to focus on multi-center clinical trials that involve a heterogeneous population in future studies to confirm generalizability. Moreover, the use of quantum-AI accelerators may enhance scalability and personalization as multimodal datasets grow in size and complexity, which is based on the progress in computational healthcare [1]. Lastly, it is essential to align the system with the global ethical AI standards that focus on the ideas of fairness, inclusivity, and accountability to be able to make the technology responsibly applicable in different cultural and healthcare contexts [14,15]. 6. Conclusion This research suggested a multi-modal federated-edge AI system of autism care involving the combination of speech, movement, and physiological cues into a safe and understandable clinical decision-support system (CDSS). Through the integration of edge intelligence to ensure real-time monitoring, federated privacy-preserving learning, and clinician- World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 397 facing explainability dashboards, the framework showed quantifiable increases in predictive performance, latency reduction and clinician trust. These findings highlight its potential as a technical and ethically sound healthcare AI model. Notably, the framework is not confined to the case of autism only. Its stratum architecture provides a scalable structure to behavioral health surveillance and workforce optimization and precision medicine, so it is applicable in various sectors of healthcare innovation. It will make the predictive insights workable and reliable in the reality of clinical workflows by introducing explainability and awareness of the workforce into the system. In perspective, future research will focus on real-world validation by multi-center clinical trials, integrating quantumAI accelerators to improve scalability and personalization, and addressing global ethical AI theories that encourage fairness, transparency, and sustainability. Combined, these instructions will help make the framework a responsible, human-focused AI solution that will allow revolutionizing the care of autism and moving the field of healthcare decision support forward. Compliance with ethical standards Disclosure of conflict of interest No conflict of interest to be disclosed. References [1] Akhi, S. S., Hussain, A. H., Islam, S., Hasan, M. N., & Islam, M. M., (2023). Artificial Intelligence–Enhanced Quantum Computing for Medical Simulations: Accelerating Drug Discovery, Protein Folding, and Personalized Healthcare with Performance Optimization and Predictive Accuracy. World Journal of Advanced Research and Review, 18(01), 1328-1335 https://doi.org/10.30574/wjarr.2023.18.1.0694 [2] Islam, M. M., Hussain, A. H., Hasan, M. N., Akhi, S. S., Hussain, Hossain, M. S., & Islam, S. (2023). Fraud detection: Develop skills in fraud detection, which is a critical area of business analytics. World Journal of Advanced Research and Review, 18(03), 1664-1672 https://doi.org/10.30574/wjarr.2023.18.3.0969 [3] Islam, M. M., & Mim, S. S. (2023). Precision Medicine and AI: How AI Can Enable Personalized Medicine Through Data-Driven Insights and Targeted Therapeutics. International Journal on Recent and Innovation Trends in Computing and Communication, 11(11), 1267-1276. https://doi.org/10.17762/ijritcc.v11i11.11359 [4] Raihena, S. M. S., Mariam, Q. R., Rashaq, M., Arif, M. A. H., & Hussain, A. H. (2023). Learning what works in accommodations: A federated machine learning framework with differential privacy guarantees. World Journal of Advanced Research and Reviews, 20(3), 2374–2379. https://doi.org/10.30574/wjarr.2023.20.3.2492 [5] Satyanarayanan M. The emergence of edge computing. Computer. 2017;50(1):30-9. doi:10.1109/MC.2017.9. [6] Islam, M. M., Arif, M. A. H. ., Hussain, A. H., Raihena, S. M. S. ., Rashaq, M. ., & Mariam, Q. R. (2023). Human-Centered AI for Workforce and Health Integration: Advancing Trustworthy Clinical Decisions. Journal of Neonatal Surgery, 12(1), 89–95. https://doi.org/10.63682/jns.v14i32S.9123 [7] Qazi Rubyya Mariam, Ariful Haque Arif, Abdullah Hill Hussain, Munadil Rashaq and S M SHAH RAIHENA. Building trustworthy clinical AI: usable checklists and transparency artifacts tested in real-world health systems. World Journal of Advanced Research and Reviews, 2025, 26(03), 2804–2810. Article DOI: https://doi.org/10.30574/wjarr.2025.26.3.2203 [8] Islam, M. M., Anika, A., Mim, S. S., Hasan, A., & Salam, S. (2024). Wearable technology: Exploring the interrogation of electronics in clothing. World Journal of Advanced Research and Reviews, 24(03), 2219-2228. https://doi.org/10.30574/wjarr.2024.24.3.3925 [9] Islam, M. M., Hussain, A. H., Mariam, Q. R., Islam, S., & Hasan, M. N. (2025). AI-Enabled predictive health monitoring for children with autism using IOT and machine learning to detect behavioral changes. Perinatal Journal, 33(1), 415-422. http://dx.doi.org/10.57239/prn.25.03310048 [10] Munadil Rashaq, Md Ariful Haque Arif, Abdullah Hill Hussain, Qazi Rubyya MariamS M Shah Raihena(2024) Preventing Overdoses, Expanding Mental Health Access, and Integrating Behavioral and Physical Health Across the 988 Continuum: An AI-Driven Approach.Frontiers in Health Informatics, (6), 4658-4667 https://doi.org/10.63682/fhi416 World Journal of Advanced Research and Reviews, 2025, 28(01), 392-398 398 [11] Ariful Haque Arif, Abdullah Hill Hussain, S M SHAH RAIHENA, Qazi Rubyya Mariam and Munadil Rashaq. Artificial Intelligence for Safer Crisis Response: De-escalation Prompts and Emergency Handoff Optimization. World Journal of Advanced Research and Reviews, 2025, 25(02), 2785–2792. Article DOI: https://doi.org/10.30574/wjarr.2025.25.2.0660. [12] Md Maruful Islam. (2024). Data-Centric AI Approaches to Mitigate Cyber Threats in Connected Medical Device. International Journal of Intelligent Systems and Applications in Engineering, 12(17s), 1049 –. Retrieved from https://ijisae.org/index.php/IJISAE/article/view/7763 [13] Shah Raihena, S. M., Haque Arif, M. A., Rubyya Mariam, Q., Hussain Munadil Rashaq, A., & Hill, A. (2025). AIEnhanced Decision Support Systems for Autism Caregivers: Redefining HR’s Role in Workforce Planning and Patient-Centered Care. Frontiers in Health Informatics, 13(2). https://doi.org/10.63682/fhi2698 [14] Floridi L, Cowls J. A Unified Framework of Five Principles for AI in Society. Harvard Data Science Review. 2019;1(1). doi:10.1162/99608f92.8cd550d1. [15] Jobin A, Ienca M, Vayena E. The global landscape of AI ethics guidelines. Nat Mach Intell. 2019;1:389–99. doi:10.1038/s42256-019-0088-2.