A Novel Machine Learning Approach to Assist Early Diagnosis of Diffuse Panbronchiolitis
Hwan Jin Lee1,2, Kyung Joon Heo3, Yeon Seok You2,4,5
1Department of Internal Medicine and Research Center for Pulmonary Disorders, Jeonbuk National University Medical School, Jeonju, Korea.
Journal of Korean Medical Science
|December 9, 2025
Summary
Machine learning improves early diagnosis of diffuse panbronchiolitis (DPB), a rare lung disease. Allergic rhinitis and CT scan macronodules are key indicators for prompt detection.
Area of Science:
- Pulmonary Medicine
- Medical Informatics
Background:
- Diffuse panbronchiolitis (DPB) is a rare, progressive inflammatory small airway disease.
- DPB is frequently misdiagnosed as other respiratory conditions like nontuberculous mycobacterial infection or bronchiectasis.
Purpose of the Study:
- To enhance early diagnostic accuracy for DPB using machine learning (ML) algorithms.
- To identify key indicators for improved DPB diagnosis.
Main Methods:
- Applied seven ML models to clinical, laboratory, and radiological data from 99 suspected DPB patients.
- Categorized patients based on established DPB diagnostic criteria.
Main Results:
- The least absolute shrinkage and selection operator regression model achieved the highest predictive accuracy.
- Identified allergic rhinitis and CT scan macronodules as significant diagnostic factors for DPB.
Conclusions:
- This study represents the first ML application for DPB diagnosis.
- Allergic rhinitis and CT macronodules are crucial for early DPB detection.
- ML integration can improve rare disease diagnosis efficiency; further validation with larger datasets is recommended.
