HeartFedGAT: An Explainable Federated Multi-scale CNN-Graph Attention Network with Protected Model-update Transmission for Heart Disease Prediction

Cardiovascular diseases continue to pose a significant worldwide health challenge, underscoring the need for reliable, privacy-conscious prediction models that support timely diagnostic workflows.Although existing methods can capture complex relationships among clinical features, achieving both model interpretability and data privacy remains challenging.To address these issues, this study presents HeartFedGAT, an Explainable Federated Multi-Scale CNN-Graph Attention Network for heart disease prediction.The framework integrates multi-scale convolutional feature extraction, a graph layer with learnable adjacency, graph attention-based feature refinement, Fernet-encrypted model-update transmission with sample-size-weighted Federated Averaging (FedAvg), and SHapley Additive exPlanations (SHAP) for model interpretability.The proposed framework was evaluated using the UCI Heart Disease dataset, comprising 3,235 patient records with 14 clinical attributes, distributed across multiple federated clients for collaborative learning.HeartFedGAT achieved a global classification accuracy of 93.35%, alongside high precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC).An ablation study and comparisons with representative baseline models provided quantitative evidence of the contributions of the major architectural components under the evaluated experimental setting.Furthermore, SHAP analysis identified influential clinical features, supporting the interpretation of model predictions.Overall, the results demonstrate the potential of HeartFedGAT as an explainable and privacy-conscious framework for heart disease prediction in a simulated federated learning setting.However, external validation is required to establish its generalizability and applicability across multiple clinical institutions.

Authors

Institutions

Publication Details

Journal
International journal of intelligent engineering and systems
Published
2026-11-17
DOI
https://doi.org/10.22266/ijies2026.1231.30
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

HeartFedGAT: An Explainable Federated Multi-scale CNN-Graph Attention Network with Protected Model-update Transmission for Heart Disease Prediction

Anagha Achuthan, A Saritha
1 links
International journal of intelligent engineering and systems
Privacy-Preserving Technologies in Data
article

HeartFedGAT: An Explainable Federated Multi-scale CNN-Graph Attention Network with Protected Model-update Transmission for Heart Disease Prediction

Anagha Achuthan, A Saritha
article en

Abstract

Cardiovascular diseases continue to pose a significant worldwide health challenge, underscoring the need for reliable, privacy-conscious prediction models that support timely diagnostic workflows.Although existing methods can capture complex relationships among clinical features, achieving both model interpretability and data privacy remains challenging.To address these issues, this study presents HeartFedGAT, an Explainable Federated Multi-Scale CNN-Graph Attention Network for heart disease prediction.The framework integrates multi-scale convolutional feature extraction, a graph layer with learnable adjacency, graph attention-based feature refinement, Fernet-encrypted model-update transmission with sample-size-weighted Federated Averaging (FedAvg), and SHapley Additive exPlanations (SHAP) for model interpretability.The proposed framework was evaluated using the UCI Heart Disease dataset, comprising 3,235 patient records with 14 clinical attributes, distributed across multiple federated clients for collaborative learning.HeartFedGAT achieved a global classification accuracy of 93.35%, alongside high precision, recall, F1-score, and area under the receiver operating characteristic curve (AUC).An ablation study and comparisons with representative baseline models provided quantitative evidence of the contributions of the major architectural components under the evaluated experimental setting.Furthermore, SHAP analysis identified influential clinical features, supporting the interpretation of model predictions.Overall, the results demonstrate the potential of HeartFedGAT as an explainable and privacy-conscious framework for heart disease prediction in a simulated federated learning setting.However, external validation is required to establish its generalizability and applicability across multiple clinical institutions.

International journal of intelligent engineering and systemsVol. 19(12)
Vels 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