Author
Listed:
- Angelin Jeba P.
- Jasmine David D.
- Selvarathi M.
- Kenneth Bryan Nesh
- Alan Biju Varkey
- Jemima Jebaseeli T.
Abstract
Coronary artery disease (CAD) remains the leading cause of death around the globe and hence requires reliable, non-invasive, and secure diagnostic methods. In order to diagnose CAD in its early stages, we present a novel architecture of a Federated Attention-Capsule Convolutional Neural Network (FAC-CNN) integrated with Bio-Inspired Optimization Algorithms (BIOA) in dealing with multiple modalities of physiological data such as Electrocardiography (ECG), Photoplethysmogram (PPG), and Blood Pressure (BP). The model is evaluated using unimodal (only ECG signals) and multimodal physiological datasets to validate the proposed algorithm for different data distributions. To improve accuracy, the architecture of FAC-CNN uses capsule networks to recognize the hierarchical structure within the data fed into it. Through attention mechanisms, the network is capable of selecting the most important features related to cardiac diseases. To ensure data security, the federated learning (FL) method is applied as a solution to the problem of local model training using patients' private data stored in edge computing devices. A novel hybrid optimization method combining Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) is used to update learning rates, the total number of capsule layers, and routing iterations. Three databases, including the PTB diagnostic ECG database, the MIT-BIH Arrhythmia database, and the MIMIC-III Waveform database, which comprises authentic ICU recordings of ECG, PPG, and blood pressure data from more than 40,000 patients were employed to evaluate the model. It can be seen from the results that FAC-CNN outperformed the other models used in the experiment, including CNNs and LSTMs by 4-7%, obtaining a mean accuracy of 97.4%, an F1-score of 96.8%, and an AUC score of 0.985. Moreover, the scalability of the proposed algorithm for applications in remote healthcare settings improved, and the more effective training method contributed to time-savings for the training procedure, decreasing it by 18.6%.
Suggested Citation
Handle:
RePEc:gvu:tcsst0:v:8:y:2026:i:2:id:2173
DOI: 10.36548/jtcsst.2026.2.001
Download full text from publisher
Corrections
All material on this site has been provided by the respective publishers and authors. You can help correct errors and omissions. When requesting a correction, please mention this item's handle: RePEc:gvu:tcsst0:v:8:y:2026:i:2:id:2173. See general information about how to correct material in RePEc.
If you have authored this item and are not yet registered with RePEc, we encourage you to do it here. This allows to link your profile to this item. It also allows you to accept potential citations to this item that we are uncertain about.
We have no bibliographic references for this item. You can help adding them by using this form .
If you know of missing items citing this one, you can help us creating those links by adding the relevant references in the same way as above, for each refering item. If you are a registered author of this item, you may also want to check the "citations" tab in your RePEc Author Service profile, as there may be some citations waiting for confirmation.
For technical questions regarding this item, or to correct its authors, title, abstract, bibliographic or download information, contact: Journal Manager (email available below). General contact details of provider: https://irojournals.com/tcsst .
Please note that corrections may take a couple of weeks to filter through
the various RePEc services.