TEA-XFEL: trustworthy and energy-aware explainable federated edge learning for secure heterogeneous healthcare intelligence

Introduction Artificial Intelligence (AI) and Internet of Medical Things (IoMT)-enabled Healthcare systems have significantly improved intelligent clinical decision-making through distributed data analytics. However, conventional centralized learning approaches remain vulnerable to patient privacy leakage, excessive communication overhead, limited transparency, adversarial attacks, and inefficient resource utilization in edge-based Healthcare environments. Although Federated Learning (FL) provides a decentralized paradigm for privacy-preserving collaborative learning, existing Healthcare FL frameworks frequently rely on static privacy configurations and provide limited integration of trust management, explainability, and resource-aware client participation. Methods This study proposes a secure and scalable Trustworthy and Energy-Aware Explainable Federated Edge Learning (TEA-XFEL) framework for distributed Healthcare intelligence across heterogeneous data modalities. The framework integrates (i) adaptive differential privacy budgeting, (ii) blockchain-enabled reputation-aware secure aggregation, (iii) energy-aware client selection, (iv) Explainable Federated Learning (XFL), and (v) lightweight knowledge-distillation-based edge inference. The framework is evaluated independently using the MIMIC-III Electronic Health Records dataset for in-hospital mortality prediction and the NIH ChestX-ray14 medical-imaging dataset for multi-label thoracic finding classification. The two datasets represent unrelated patient populations; therefore, no patient-level pairing, cross-dataset feature concatenation, or multimodal fusion is performed. For each dataset, the training data are distributed across 20 federated Healthcare clients under a non-IID setting to emulate heterogeneous cross-institutional environments. Results Experimental evaluation demonstrates the applicability of TEA-XFEL to both structured EHR and medical-imaging tasks, while the proposed energy-aware mechanism reduces communication overhead by 38.7% and energy consumption by 31.4%. The integrated XAI component provides modality-specific computational explanations using appropriate attribution mechanisms for structured clinical data and medical images. Discussion Overall, TEA-XFEL provides a privacy-aware, trustworthy, explainable, and resource-efficient federated architecture for heterogeneous Healthcare data while maintaining independent modality-specific learning and evaluation.

Authors

Institutions

Publication Details

Journal
Frontiers in Artificial Intelligence
Published
2026-11-17
DOI
https://doi.org/10.3389/frai.2026.1939671
Primary Topic
Privacy-Preserving Technologies in Data
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

TEA-XFEL: trustworthy and energy-aware explainable federated edge learning for secure heterogeneous healthcare intelligence

K R Jothi, Munusamy S
3 links
Frontiers in Artificial Intelligence
Privacy-Preserving Technologies in Data
article

TEA-XFEL: trustworthy and energy-aware explainable federated edge learning for secure heterogeneous healthcare intelligence

K R Jothi, Munusamy S
article en

Abstract

Introduction Artificial Intelligence (AI) and Internet of Medical Things (IoMT)-enabled Healthcare systems have significantly improved intelligent clinical decision-making through distributed data analytics. However, conventional centralized learning approaches remain vulnerable to patient privacy leakage, excessive communication overhead, limited transparency, adversarial attacks, and inefficient resource utilization in edge-based Healthcare environments. Although Federated Learning (FL) provides a decentralized paradigm for privacy-preserving collaborative learning, existing Healthcare FL frameworks frequently rely on static privacy configurations and provide limited integration of trust management, explainability, and resource-aware client participation. Methods This study proposes a secure and scalable Trustworthy and Energy-Aware Explainable Federated Edge Learning (TEA-XFEL) framework for distributed Healthcare intelligence across heterogeneous data modalities. The framework integrates (i) adaptive differential privacy budgeting, (ii) blockchain-enabled reputation-aware secure aggregation, (iii) energy-aware client selection, (iv) Explainable Federated Learning (XFL), and (v) lightweight knowledge-distillation-based edge inference. The framework is evaluated independently using the MIMIC-III Electronic Health Records dataset for in-hospital mortality prediction and the NIH ChestX-ray14 medical-imaging dataset for multi-label thoracic finding classification. The two datasets represent unrelated patient populations; therefore, no patient-level pairing, cross-dataset feature concatenation, or multimodal fusion is performed. For each dataset, the training data are distributed across 20 federated Healthcare clients under a non-IID setting to emulate heterogeneous cross-institutional environments. Results Experimental evaluation demonstrates the applicability of TEA-XFEL to both structured EHR and medical-imaging tasks, while the proposed energy-aware mechanism reduces communication overhead by 38.7% and energy consumption by 31.4%. The integrated XAI component provides modality-specific computational explanations using appropriate attribution mechanisms for structured clinical data and medical images. Discussion Overall, TEA-XFEL provides a privacy-aware, trustworthy, explainable, and resource-efficient federated architecture for heterogeneous Healthcare data while maintaining independent modality-specific learning and evaluation.

Frontiers in Artificial IntelligenceVol. 9
Vellore Institute of Technology University (IN)
Good health and well-being
Privacy-Preserving Technologies in Data
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon