13 November 2024 in Expert Systems
16 December 2025 in IEEE Access
Eye-tracking technology provides an estimate of where a person is looking and is essential for understanding individual attention and the psychological functions within the cognitive system. It can aid in understanding neurodevelopmental disorders and tracing a person’s identity. However, this technology poses a significant privacy risk, as it captures sensitive information about individuals and increases the likelihood that their data can be traced back to them. This paper proposes a human-centered two-phased framework to prevent identity backtracking while preserving the pedagogical benefits of AI-powered eye-tracking in interactive learning environments. We first demonstrate the potential for backtracking student IDs and diagnoses in various scenarios using serious game-based eye-tracking data. We then provide a two-stage privacy-preserving framework that prevents participants from being tracked while still enabling diagnostic classification. The first phase covers four scenarios: I) predicting disorder diagnoses based on different game levels, II) predicting student IDs based on different game levels, III) predicting student IDs based on randomized data, and IV) utilizing K-Means for out-of-sample data. In the second phase, we present a two-phase framework that preserves privacy. We also employ Federated Learning (FL) across multiple clients, incorporating a secure identity management system with dummy IDs and administrator-only access controls. In the first phase, the proposed framework achieved 99.3% accuracy in scenario 1, 63% in scenario 2, and 99.7% in scenario 3, and successfully identified and assigned a new student ID in scenario 4. In phase 2, we effectively prevented backtracking and established a secure identity management system with dummy IDs and administrator-only access controls, achieving an overall accuracy of 99.40%. Our dual-layer anonymization scheme aligns with the GDPR’s principles of data minimization and purpose limitation, providing a foundational privacy protection mechanism for federated learning in educational settings. While it does not ensure full regulatory compliance, our approach illustrates how federated architectures can incorporate privacy-by-design principles to limit re-identification risks. This proof-of-concept study demonstrates the architectural feasibility of our framework, achieving accuracy rates ranging from 63.03% to 99.73% across various scenarios. However, the extreme variability in performance and a small sample size (n=9) suggest significant overfitting. Consequently, the findings need large-scale validation to confirm their generalizability.
3 February 2026 in IEEE Access
This paper introduces an innovative adaptive scoring framework for children with Neurodevelopmental Disorders (NDD) that integrates multiple metrics, such as spatial attention patterns, temporal engagement, and game performance data, to provide a comprehensive assessment of learning that extends beyond traditional game scoring. The framework employs a progressive difficulty adaptation method that focuses on specific stimuli at each level and dynamically adjusts weights to accommodate increasing cognitive load and learning complexity. Additionally, it includes capabilities for temporal analysis, such as detecting engagement periods, rewarding sustained attention, and implementing an adaptive multiplier framework based on performance levels. To avoid over-rewarding high performers while maximizing improvement potential for struggling students, the designed framework features an adaptive temporal impact framework that adjusts performance scales accordingly. We also established a multi-metric validation framework using Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Pearson correlation, and Spearman correlation, along with defined quality thresholds, to assess deployment readiness in educational settings. This research bridges the gap between technical eye-tracking metrics and educational insights by explicitly mapping attention patterns to learning behaviors, enabling actionable pedagogical interventions.