Development and application of a deep learning-based comprehensive early diagnostic model for chronic obstructive...

Area of Science:

• Medical imaging analysis
• Artificial intelligence in healthcare
• Pulmonary medicine

Background:

• Chronic obstructive pulmonary disease (COPD) requires early diagnosis and effective management for better patient outcomes.
• Current diagnostic approaches can be enhanced through advanced computational methods.

Purpose of the Study:

• To develop and evaluate an advanced COPD diagnostic model.
• To integrate deep learning, radiomics, and epidemiological data for improved diagnostic accuracy.

Main Methods:

• Utilized CT images and questionnaire data from 2,983 participants.
• Extracted deep learning features using Variational Autoencoder and radiomics features using PyRadiomics.
• Developed standalone, fusion, and comprehensive models using Multi-Layer Perceptrons.

Main Results:

• The fusion model (deep learning + radiomics) achieved an AUC of 0.952.
• Standalone models showed lower performance: deep learning (AUC=0.844) and radiomics (AUC=0.944).
• The comprehensive model integrating all data sources yielded the highest AUC of 0.971.

Conclusions:

• A data fusion strategy effectively created a state-of-the-art COPD diagnostic model.
• The comprehensive model demonstrates superior diagnostic performance and clinical applicability.