Jul 2026· 2026 4th International Conference on Sustainable Computing and Smart Systems (ICSCSS)· pp. 1411-1416· 0 citations· 10 references
Abstract
Deep learning based medical diagnosis systems are often hindered by privacy restrictions on patients’ health data. In this paper, we propose a privacy preserving decentralized algorithm for multi-label classification of chest diseases using Federated Learning (FL). We experimented on 26,218 images sampled from NIH Chest X-ray dataset with a ResNeSt-50 deep neural network to classify 8 thoracic abnormalities. ResNeSt-50 incorporates Split-Attention mechanisms, which improve multiscale feature extraction for chest disease classification. As chest x-rays are low contrast images by nature, we applied Contrast Limited Adaptive Histogram Equalization (CLAHE) preprocessing to input images. The experimental results show that our proposed FL method obtains a competitive accuracy score of 7.04/8.0 (88.01%) compared to the non-decentralized baseline with the added benefits of FL, such as reduction in validation loss. This work aims to serve as a baseline for future work in FL for chest imaging.
The classification of respiratory sounds remains a challenging task in clinical research, limited by severe dataset imbalance and strict privacy regulations. Traditional machine learning methods are centralized and jeopardize data privacy and have difficulty in identifying minority classes, which makes them unsuitable for real-world clinical use.
In this paper, a novel federated learning framework for multi-class respiratory disease classification is proposed based on the fusion of a Hybrid Support Vector Machine-K-Nearest Neighbor (SVMKNN) classifier with Adaptive Federated Aggregation (AFA), Synthetic Minority Oversampling Technique (SMOTE) balancing, and a Gaussian differential privacy (DP) mechanism. The suggested framework runs using five non-independently and identically distributed (non-IID) client nodes partitioned using Dirichlet distribution (α = 1.0) and able to perform privacy-preserving collaborative training without sharing raw data. We extract a rich multi-modal acoustic feature representation from respiratory recordings using Mel-Frequency Cepstral Coefficients (MFCCs), first and second order delta-MFCCs, Mel spectrograms, chroma features, spectral contrast, and a set of seven scalar acoustic descriptors. The resulting high-dimensional feature vector is compressed using Principal Component Analysis (PCA) to retain 95% cumulative variance. The hybrid classifier combines the discriminative boundary learning of SVM with the local density estimation of KNN through a weighted probability ensemble (α = 0.6), resulting in better generalization in federated, class-imbalanced settings. A major methodological innovation is the restriction of the differential privacy noise to the aggregation phase, so that all reported metrics are based on clean, noise-free model outputs.
The suggested framework obtains an accuracy of 97.55, a weighted F1-score of 97.64, Matthews Correlation Coefficient (MCC) of 90.89, and Cohen's Kappa of 90.77.
Ablation studies confirm that federated hybrid learning with AFA and SMOTE outperforms standalone SVM, standalone KNN, and centralized hybrid baselines for all evaluation metrics, proving the clinical viability of the proposed framework for distributed hospital networks and telemedicine platforms.
Syed Riyazul Haq, K. Lakshmanna· Frontiers in Digital Health· 0 citations
FedGI-Screen demonstrates that privacy-preserving FL can match or exceed the performance of centralised models for GI disease screening, while maintaining rigorous data confidentiality compliance with GDPR and HIPAA.
S. Nithiya, S. Murugaanandam, K. Sornalakshmi et al.· International Journal of Onl...· 0 citations
ABSTRACT Skin cancer is among the most common malignant tumors worldwide, and early detection is essential to improve patient survival and recovery. Conventional AI‐based diagnostic approaches rely on centralized data, which raises security and privacy concerns. This paper presents an implementation of federated learning that addresses these challenges by enabling collaborative model training among simulated distributed client nodes while ensuring that raw patient data remains confidential. A convolutional neural network architecture, ResNet50, is employed at client nodes, with preprocessing steps including image augmentation, contrast enhancement, and lesion segmentation to improve feature extraction on the International Skin Imaging Collaboration (ISIC) dataset. Multiple aggregation algorithms are implemented at the central server, along with both vertical and horizontal federated learning settings. Their performance is evaluated using metrics such as precision, accuracy, recall, and F1‐score. In the horizontal federated learning setting, the FedNova aggregation approach outperformed other methods, achieving an accuracy of 76.45%, F1‐score of 0.73, and a recall of 0.72, demonstrating enhanced overall classification performance among the evaluated aggregation algorithms and stable convergence across clients. These results highlight the impact of federated learning settings and aggregation strategies on overall skin cancer detection performance.
B. Shrimali, Rebekah Geddam, H. Ghayvat et al.· Healthcare technology letter...· 0 citations
Timely analysis of cardiac MRI data is essential for early diagnosis and clinical intervention in cardiovascular illness, which continues to be the one world's leading causes of death. However, cross-institutional data sharing is ethically and legally problematic due to the nature of sensitive medical imaging data, which restricts the capacity to train reliable diagnosis models on the variety of the patient groups. A federated DL system for privacy-preserving anomaly detection and classification in cardiac MRI is presented in this work. It is intended to function in various simulated hospital settings without disclosing raw patient data. The suggested system, HeartMRI-FL, combines two deep learning models, an adaptive sub-client splitting mechanism, client-level differential privacy, and a Flower-based federated learning architecture— a supervised classifier for categorizing cardiac conditions and a convolutional autoencoder for unsupervised anomaly detection. Real-time training triggers, prediction queries, and result visualization are provided by a supplementary Angular-based clinical dashboard that interacts with the federated backend over a REST API. The system not only supports popular medical imaging formats like DICOM, NIfTI, and JPEG, but it also demonstrates that substantial diagnostic performance may be achieved without centralizing patient data. The federated approach provides a workable route toward privacy-compliant multi-institutional cardiac imaging analysis by achieving competitive accuracy while upholding formal differential privacy commitments, as confirmed by experimental results.
Chaithanya L, Veena K· International Research Journ...· 0 citations
Cardiovascular disease remains a leading cause of mortality worldwide, posing challenges for early detection within privacy-sensitive medical environments. Federated Learning (FL) offers a promising solution by enabling distributed model training without transferring raw patient data to a central server. However, FL deployment on edge devices is constrained by limited resources, system heterogeneity, and non-IID data, which degrade model stability and convergence. This study integrates sparsification-based compression and 16-bit quantization into FedProx and evaluates their combined and individual effects on heart disease prediction under compounded non-IID and straggler conditions. Using the Heart Disease Train-Test dataset from the UCI repository (1,025 samples, 14 attributes) and a KDD methodology, experiments simulated a non-IID FL environment with 10 clients and 30% stragglers over 50 rounds, averaged over five seeds. Results show that standard FedProx achieves the highest mean performance (86.44% accuracy, 87.34% F1-Score), closely followed by FedAvg (86.15%, 87.18%), while the proposed method attains a lower mean accuracy (84.88%) and F1-Score (85.87%), not statistically significant (p = 0.061, 0.098). ROC-AUC, however, is significantly lower for the proposed method (93.52% vs. 94.42%, p < 0.01), indicating reduced discriminative ability. An ablation study shows compression, not quantization, primarily drives the efficiency gains, reducing transmitted parameter size by 45.09% and accelerating convergence to 20.2 rounds compared with the baseline methods. These findings indicate that integrating compression and quantization into FedProx achieves substantial communication efficiency and faster convergence with only minor performance trade-offs, offering a viable option for resource-constrained edge deployments.
The precise identification of thoracic disorders using chest X-rays (CXRs) is essential for favorable clinical diagnosis. However, it continues to be difficult because of inter-class similarities and overlapping symptoms. This study introduces a hybrid MobileNetV2–Vision Transformer (ViT) framework for multi-label classification of CXR images into 14 disease categories on NIH CXR14 dataset. MobileNetV2 serves as a lightweight feature extractor to obtain distinctive spatial representations that are tokenized and enhanced with positional encodings. The encoded features were processed via transformer layers, wherein self-attention mechanisms captured long-range interdependence and global contextual relationships. An output layer based on the sigmoid function facilitated the concurrent prediction of various illness labels. The novelty of the proposed work lies in the PSO-tuned hybrid architecture, which adaptively optimizes key hyperparameters of the MobileNetV2–ViT framework to achieve improved accuracy, faster convergence, and enhanced computational efficiency compared to conventional approaches. This implementation uses an economical, GPU-accelerated instance tailored for inference workloads. It is equipped with a single NVIDIA A10G GPU with 24 GB of dedicated GPU memory, and the proposed work is carried out on AWS SageMaker. The experimental assessment employing five-fold cross-validation showed strong performance, with an average accuracy of 94.6%, precision of 88.1%, recall of 85.9%, specificity of 94.17%, and F1-score of 91.1%. Grad-CAM representations increased interpretability by identifying disease-relevant areas. The results underscore the efficacy of the proposed hybrid technique as a dependable instrument for computer-aided diagnosis of thoracic illnesses.
R. Raj, Pavan M. P. Kumar, K. N. Manjunath et al.· Scientific Reports· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
