Using Machine Learning with Impulse Oscillometry Data to Develop a Predictive Model for Chronic Obstructive Pulmonary...

Area of Science:

• Pulmonary Medicine
• Biomedical Engineering
• Data Science

Background:

• Impulse oscillometry system (IOS) provides valuable data for respiratory assessment.
• Accurate differentiation between asthma and chronic obstructive pulmonary disease (COPD) is clinically significant.
• Machine learning offers potential for analyzing complex physiological datasets.

Purpose of the Study:

• To develop and validate a machine learning model for classifying patients into three categories: no airflow obstruction, asthma, or COPD, using IOS profiles.
• To determine the efficacy of machine learning in distinguishing between obstructive lung diseases based solely on IOS data.
• To identify optimal IOS parameters and machine learning algorithms for accurate classification.

Main Methods:

• Utilized a dataset of 480 patients (240 with COPD, 240 with asthma) and 84 healthy individuals for model training.
• Combined physiological and IOS parameters into six feature combinations.
• Evaluated five classification algorithms: logistic regression, random forest, neural network, k-nearest neighbor, and support vector machine.

Main Results:

• The neural network classifier achieved the highest accuracy (0.786) for classifying individuals into no obstruction, asthma, or COPD categories.
• Optimal classification for distinguishing healthy from unhealthy individuals involved 23 features and the neural network (accuracy 0.929).
• Differentiating between COPD and asthma yielded the best results with 15 features using the neural network (accuracy 0.854).

Conclusions:

• Machine learning models utilizing IOS profiles can accurately classify obstructive lung diseases.
• The neural network algorithm demonstrated superior performance in discriminating between healthy individuals, asthma, and COPD.
• IOS data-driven models show promise for improving the management of COPD and asthma.