Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Federated Learning for Pediatric Pneumonia Detection: Enabling Collaborative Diagnosis Without Sharing Patient Data

Daniel M. Jimenez-Gutierrez, Enrique Zuazua, Joaquin Del Rio, Oleksii Sliusarenko, Xabi Uribe-Etxebarria

TL;DR

This work addresses the challenge of diagnosing pediatric pneumonia from chest X-rays in a privacy-conscious setting where data are distributed across hospitals. It evaluates Horizontal Federated Learning using the Sherpa.ai platform to train a pneumonia classifier without sharing raw images, simulating non-IID distributions across five nodes with Dirichlet partitioning. The federated model achieves $0.900$ accuracy and a $0.966$ ROC-AUC, delivering substantial gains over single-hospital models and closely matching the centralized baseline (gains of about $47.5\%$ in accuracy and $50.0\%$ in ROC-AUC over single-site performance). The findings demonstrate that FL can provide high-performing, privacy-preserving, and scalable collaborative AI for pediatric pneumonia detection, with important implications for rare diseases and data-siloed clinical environments.

Abstract

Early and accurate pneumonia detection from chest X-rays (CXRs) is clinically critical to expedite treatment and isolation, reduce complications, and curb unnecessary antibiotic use. Although artificial intelligence (AI) substantially improves CXR-based detection, development is hindered by globally distributed data, high inter-hospital variability, and strict privacy regulations (e.g., HIPAA, GDPR) that make centralization impractical. These constraints are compounded by heterogeneous imaging protocols, uneven data availability, and the costs of transferring large medical images across geographically dispersed sites. In this paper, we evaluate Federated Learning (FL) using the Sherpa.ai FL platform, enabling multiple hospitals (nodes) to collaboratively train a CXR classifier for pneumonia while keeping data in place and private. Using the Pediatric Pneumonia Chest X-ray dataset, we simulate cross-hospital collaboration with non-independent and non-identically distributed (non-IID) data, reproducing real-world variability across institutions and jurisdictions. Our experiments demonstrate that collaborative and privacy-preserving training across multiple hospitals via FL led to a dramatic performance improvement achieving 0.900 Accuracy and 0.966 ROC-AUC, corresponding to 47.5% and 50.0% gains over single-hospital models (0.610; 0.644), without transferring any patient CXR. These results indicate that FL delivers high-performing, generalizable, secure and private pneumonia detection across healthcare networks, with data kept local. This is especially relevant for rare diseases, where FL enables secure multi-institutional collaboration without data movement, representing a breakthrough for accelerating diagnosis and treatment development in low-data domains.

Federated Learning for Pediatric Pneumonia Detection: Enabling Collaborative Diagnosis Without Sharing Patient Data

TL;DR

This work addresses the challenge of diagnosing pediatric pneumonia from chest X-rays in a privacy-conscious setting where data are distributed across hospitals. It evaluates Horizontal Federated Learning using the Sherpa.ai platform to train a pneumonia classifier without sharing raw images, simulating non-IID distributions across five nodes with Dirichlet partitioning. The federated model achieves accuracy and a ROC-AUC, delivering substantial gains over single-hospital models and closely matching the centralized baseline (gains of about in accuracy and in ROC-AUC over single-site performance). The findings demonstrate that FL can provide high-performing, privacy-preserving, and scalable collaborative AI for pediatric pneumonia detection, with important implications for rare diseases and data-siloed clinical environments.

Abstract

Early and accurate pneumonia detection from chest X-rays (CXRs) is clinically critical to expedite treatment and isolation, reduce complications, and curb unnecessary antibiotic use. Although artificial intelligence (AI) substantially improves CXR-based detection, development is hindered by globally distributed data, high inter-hospital variability, and strict privacy regulations (e.g., HIPAA, GDPR) that make centralization impractical. These constraints are compounded by heterogeneous imaging protocols, uneven data availability, and the costs of transferring large medical images across geographically dispersed sites. In this paper, we evaluate Federated Learning (FL) using the Sherpa.ai FL platform, enabling multiple hospitals (nodes) to collaboratively train a CXR classifier for pneumonia while keeping data in place and private. Using the Pediatric Pneumonia Chest X-ray dataset, we simulate cross-hospital collaboration with non-independent and non-identically distributed (non-IID) data, reproducing real-world variability across institutions and jurisdictions. Our experiments demonstrate that collaborative and privacy-preserving training across multiple hospitals via FL led to a dramatic performance improvement achieving 0.900 Accuracy and 0.966 ROC-AUC, corresponding to 47.5% and 50.0% gains over single-hospital models (0.610; 0.644), without transferring any patient CXR. These results indicate that FL delivers high-performing, generalizable, secure and private pneumonia detection across healthcare networks, with data kept local. This is especially relevant for rare diseases, where FL enables secure multi-institutional collaboration without data movement, representing a breakthrough for accelerating diagnosis and treatment development in low-data domains.

Paper Structure

This paper contains 28 sections, 5 equations, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. Pneumonia Detection from CXRs
  4. Related Work
  5. Problem Description
  6. Output Structure.
  7. Objective.
  8. Loss Computation.
  9. Privacy-Preserving ML Solution
  10. Classical ML Approach
  11. Privacy and Regulatory Constraints
  12. Introduction to FL
  13. FL Paradigms
  14. HFL
  15. Centralized Dataset and Preprocessing
  16. ...and 13 more sections

Figures (5)

  • Figure 1: Global map illustrating hospitals joining the FL initiative to enhance model generalizability.
  • Figure 2: Example pediatric CXR images: (left) normal, (center) viral pneumonia, (right) bacterial pneumonia.
  • Figure 3: Proposed architecture for centralized training.
  • Figure 4: Proposed architecture for federated training with privacy-preserving orchestration by https://www.sherpa.ai/.
  • Figure 5: Mean metric values comparing Single-Hospital, Centralized, and Federated training settings.