Related Experiment Video
Updated: Jun 18, 2025

Generation of a Chronic Obstructive Pulmonary Disease Model in Mice by Repeated Ozone Exposure
Published on: August 25, 2017
Practice of distributed machine learning in clinical modeling for chronic obstructive pulmonary disease
Junfeng Peng1, Xujiang Liu1, Ziwei Cai1
1Department of Computer Science and Engineering, Guangdong University of Education, Guangzhou 510303, China.
This study introduces COPD average federated learning (COPD_AVG_FL) to integrate scattered patient data for better chronic obstructive pulmonary disease (COPD) diagnosis. The system significantly improves diagnostic accuracy while protecting patient privacy.
Area of Science:
- Digital health
- Medical informatics
- Machine learning in healthcare
Background:
- Chronic obstructive pulmonary disease (COPD) presents challenges due to high prevalence, morbidity, and heterogeneous data scattered across medical units.
- Integrating this dispersed data is costly and complicated by patient privacy regulations, leading to data silos.
- These data islands hinder comprehensive analysis and the development of advanced digital health solutions for COPD.
Purpose of the Study:
- To develop a distributed system for diagnosing chronic obstructive pulmonary disease (COPD) that overcomes data fragmentation and privacy concerns.
- To enable high-quality data integration and modeling for COPD using federated learning principles.
- To promote the advancement of digital health applications in managing COPD.
Main Methods:
- Collected and pre-processed real-world clinical data from COPD patients, addressing data quality issues.
- Designed and implemented a federated learning architecture named COPD average federated learning (COPD_AVG_FL) utilizing the FedAvg algorithm.
- Developed a Centralized Machine Learning (CML) model for comparative evaluation of the COPD_AVG_FL system.
Main Results:
- The COPD_AVG_FL system demonstrated substantial improvements in diagnostic performance on test data, with absolute increases of 13.4% in accuracy, 13.3% in precision, 12.8% in recall, 13.1% in F1-Score, and 12.9% in AUC.
- The federated learning approach successfully decoupled model training from raw patient data, ensuring robust privacy protection.
- The system facilitates secure integration of diverse COPD data, leading to a more comprehensive and accurate diagnostic model.
Conclusions:
- COPD_AVG_FL effectively addresses the challenges of scattered data and patient privacy in COPD diagnosis.
- The system enhances diagnostic metrics significantly, promoting the practical application of AI in clinical settings for COPD.
- This approach paves the way for secure and effective digital health solutions in managing chronic respiratory diseases.
Related Concept Videos
Chronic Obstructive Pulmonary Disease
Smoking is a primary risk factor for COPD, with over 80% of patients having a history of it. Patients typically experience progressive dyspnea or labored breathing, frequent coughing, and recurrent pulmonary infections. Many eventually succumb to respiratory failure, characterized by...
Chronic Obstructive Pulmonary Disease-V: Management
Smoking Cessation
Chronic Obstructive Pulmonary Disease-II: Pathophysiology
Chronic Inflammation
Chronic Obstructive Pulmonary Disease-I: Introduction
Chronic Obstructive Pulmonary Disease-IV: Assessement and Diagnostic Studies
Medical History
Chronic Obstructive Pulmonary Disease-V: Nursing Management
Assessment

