Reproduce “Early Diagnosis of Chronic Obstructive Pulmonary Disease from Chest X-Rays using Transfer Learning and Fusion Strategies” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Ryan Wang, Li-Ching Chen, Lama Moukheiber, Mira Moukheiber, Dana Moukheiber, Zach Zaiman, Sulaiman Moukheiber, Tess Litchman and 6 more
Researchers from National Tsing Hua University, MIT, Beth Israel Deaconess Medical Center, and Emory University developed deep learning models for early-stage Chronic Obstructive Pulmonary Disease (COPD) detection from Chest X-rays (CXRs), incorporating transfer learning and various fusion strategies. The weighted average bagging model achieved an AUC of 0.77 and F1-score of 0.54 on internal test data, while multi-site data fusion improved generalizability, reaching an AUC of 0.76 on an external dataset, demonstrating potential for an accessible and fair screening tool.
Key insights
- The Xception model consistently performed best among the base CNN architectures, showing good generalizability across internal and external datasets.
- Ensemble methods, particularly weighted average bagging, provided a modest performance improvement over single models and demonstrated better overall fairness across demographic groups.
- Multi-site data fusion significantly improved model generalizability to an external dataset, underscoring the importance of diverse data for robust AI applications in medicine.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.