Machine Learning-Assisted Portable Electrical Impedance Tomography Enables Accurate Lung Function Assessment.

Area of Science:

• Biomedical Engineering
• Pulmonary Medicine
• Medical Imaging

Background:

• Respiratory diseases cause millions of deaths globally, with limited affordable screening tools.
• Current spirometry lacks regional localization crucial for personalized respiratory care.
• Electrical impedance tomography (EIT) is promising but traditionally requires bulky equipment.

Purpose of the Study:

• To develop and evaluate a novel portable EIT system for lung function assessment.
• To integrate machine learning for enhanced respiratory monitoring and diagnosis.
• To provide an affordable, non-invasive alternative for global and regional lung assessment.

Main Methods:

• Development of a portable EIT system.
• Integration of machine learning algorithms for data analysis.
• Correlation of EIT-based lung function with standard pulmonary function tests.

Main Results:

• The portable EIT system showed good correlation with pulmonary function tests (R² values ranging from 0.538 to 0.646).
• The system demonstrated potential in diagnosing COPD with 82% sensitivity and 73% specificity.
• Machine learning provided interpretability for global and regional lung function prediction.

Conclusions:

• A portable EIT system with machine learning is a viable, affordable tool for respiratory disease monitoring.
• This technology enables both global and regional lung assessment, improving personalized care.
• Clinical relevance includes home and community-based screening for conditions like asthma and COPD.