A study on chronic obstructive pulmonary disease diagnosis using multilayer neural networks

Area of Science:

• Medical Informatics
• Artificial Intelligence in Medicine
• Pulmonology

Background:

• Chronic Obstructive Pulmonary Disease (COPD) is a progressive lung disease characterized by irreversible airflow limitation.
• COPD is a significant public health issue, particularly prevalent in Turkey.
• Cigarette smoking is the primary risk factor, with genetic factors also influencing susceptibility.

Purpose of the Study:

• To investigate the application of multilayer neural networks (MLNN) for the diagnosis of COPD.
• To evaluate the performance of different MLNN architectures in COPD detection.

Main Methods:

• Utilized two distinct MLNN structures: one with a single hidden layer and another with two hidden layers.
• Employed backpropagation with momentum and Levenberg-Marquardt algorithms for neural network training.
• Compiled a COPD dataset from patient epicrisis reports at a chest diseases hospital.

Main Results:

• The study successfully implemented MLNN for COPD diagnosis.
• Both single and double hidden layer MLNN models demonstrated capability in identifying COPD patients from clinical data.

Conclusions:

• Multilayer neural networks show promise as an effective tool for COPD diagnosis.
• This machine learning approach offers a potential advancement in the diagnostic capabilities for Chronic Obstructive Pulmonary Disease.