67471 research outputs found
Sort by
Optimizing Mental Health Referral Workflows: A Framework for Trust-Critical Decision Points
Get PDFMental health referral systems exhibit significant workflow failures, with rejection rates varying from 33% nationally to 70-90% among individual practitioners. Through systematic literature review and semi-structured interviews with nine Norwegian general practitioners, we identified five trust-critical decision points where uncertainty triggers defensive behaviors: severity assessment, confidence determination, rejection response, pathway selection, and information handoff. We developed a multi-criteria optimization framework proposing uncertainty visualization, predictive analytics, and adaptive documentation interventions. Based on comparable implementations achieving 20-40% efficiency gains, our framework addresses the gap between system metrics and practitioner reality, reconceptualizing trust as a dynamic workflow factor
Process Mining IoT-Enriched Tyre Lifecycles for Predictive Maintenance Across Fleet Operators
Get PDFTyre maintenance remains a cost-critical process for trailer fleets, yet the industry’s tread-depth rule often prompts premature replacements. We propose an IoT-aware process-mining bridge that fuses build records, workshop events, and 15-min telematics into 7 826 tyre lifecycles (> 60 000 events). Gradient-boosted trees trained on 25 lifecycle-level stress features reach R² = 0.60 and MAE ≈ 7 900 km on a 2024/25 hold-out set. Inverse-power learning curves from 10 %, 20 %, … to 100 % training slices (422 → 4 226 lifecycles) reveal a sharp elbow: accuracy plateaus after ≈ 2 500 lifecycles (60 %), and the remaining data trim MAE by only 22 km (< 0.3 %). The study quantifies data-efficiency thresholds and offers actionable benchmarks for process-technology adoption in predictive tyre maintenance. It also demonstrates how IoT-enriched event logs can be leveraged within business-process-technology frameworks to support continuous improvement and regulatory-compliance assurance
Adoption of AI-Enabled Decision Support Systems for Supply Chain Resilience – An Individual-Level Perspective
Get PDFIn the context of supply chain management, Artificial Intelligence (AI) is contributing to increasingly performing Decision Support Systems (DSSs) that can support data-driven decision-making, leading to better supply chain resilience. As this is contingent on decision-makers adopting and using such systems, it remains unclear how various technology and individual-level factors interact to influence the adoption and usage of AI-enabled DSSs in the context of supply chain management. Using a 2 (AI vs. no AI) x 2 (confirmation vs. disconfirmation) scenario-based experiment, the study finds that employees with prior experience using AI exhibit a higher intention to adopt and use an AI-enabled DSS when it generally contradicts their opinion of the best course of action. Less experienced employees tend to prefer a non-AI-enabled DSS. Under the confirmation condition, and possibly fearing overreliance on the DSS, employees with more experience using AI prefer a non-AI-driven DSS
Introduction to the Minitrack on Bright and Dark Side of Social Media in Marginalized Contexts
No full textPredicting Engagement in Human-Robot Teams via Node-Edge Co-Attention Dynamic Graph Neural Networks
Get PDFHuman-AI collaboration increasingly depends on intelligent agents that can understand and influence human social behavior. In small-group settings, conversational dynamics shape team cohesion and task success. However, existing models in dynamic graph learning struggle with small-scale graphs, high-dimensional edge features, and multi-level predictions. This paper proposes a novel Node-Edge Co-Attention Dynamic Graph Neural Network (DyNEA) for engagement prediction in human-robot teams. Using data from 30 rounds of collaborative games involving conversations from participants, our model jointly learns node, edge, and graph-level representations and make predictions. DyNEA outperforms baselines across multiple metrics. Our framework offers potential applications in human-AI collaboration, emotional support systems, and modeling cooperation dynamics
Lightweight and Privacy-Enhanced Detection Model on Aerial Imagery for Post-Disaster Building Damage Reconnaissance
Get PDFAs post-disaster aerial imagery becomes a crucial resource for structural damage assessment, automated detection systems must address challenges in classification granularity, data privacy, and deployment efficiency. To tackle these issues, we propose a lightweight and privacy-enhanced building damage detection framework that integrates YOLO-based object detection with differentially private training and structured pruning. Specifically, we apply Differentially Private Stochastic Gradient Descent (DP-SGD) to inject calibrated Laplace noise during training, offering formal -differential privacy guarantees for sensitive imagery. To enable real-time inference on edge-constrained platforms like UAVs, we further employ structured channel pruning to eliminate redundant parameters without modifying the model architecture. Empirical results on a well-annotated dataset demonstrate that our method maintains strong detection performance while achieving both privacy protection and model compactness, providing a lightweight and secure solution for timely post-disaster building damage assessment
Agentic AI Readiness: A Process-Oriented Assessment Framework
Get PDFAre organizations ready for agentic artificial intelligence (AI)? Based on the limited success of AI projects in practice, this paper proposes an approach that recognizes the new potential of agentic AI for automating business processes and posits that business benefits are created at the level of business processes. It presents a framework for evaluating organizational readiness for agentic AI systems that emphasizes their potential of agentic AI systems for business process management. The framework provides a dual assessment: first, it determines the potential readiness across five perspectives (activities, decisions, data operations, control flow, and resource management). Second, it measures process debt, which refers to the gaps between documented and actual practices. Validated through a case study involving 40+ stakeholders and nine processes at a German university, the framework reveals distinct readiness patterns and actionable transformation insights. This enables evidence-based decision-making regarding AI investments and systematic capability building for agentic AI implementation in existing organizational settings