To propose a federated learning framework for collaborative ASD screening across different age groups while emphasizing the critical importance of preserving patient privacy.
Key Findings:
Federated learning approaches achieved global accuracy of 97.2% for children, 89.5% for adolescents, and 86.8% for adults, indicating strong performance across age groups.
The proposed framework improved fairness and robustness compared to centralized models, suggesting a more equitable approach to ASD screening.
Maintained computational and communication efficiency while ensuring privacy, highlighting the practical applicability of the framework.
Interpretation:
Personalized federated learning offers a scalable and accurate solution for ASD screening across diverse age groups, integrating advanced ML techniques with ethical clinical practices that prioritize patient privacy.
Limitations:
The study does not replace the need for in-depth clinical assessments by qualified professionals, emphasizing the importance of human oversight.
Results may vary based on the quality and diversity of datasets used, which can significantly impact the generalizability of the findings.
Conclusion:
The federated learning framework supports responsible ASD detection in real-world healthcare settings, bridging the gap between technology and ethical considerations, and promoting equitable access to screening.
