Asthma and Chronic Obstructive Pulmonary Disease (COPD) Prediction Using PEEP Expiratory Occlusion Data: A Hybrid Approach with Random Forest and Feedforward Neural Network

Asthma and Chronic Obstructive Pulmonary Disease (COPD) are chronic respiratory conditions affecting millions globally, underscoring the need for early and accurate diagnosis to mitigate severe health complications. This study explores the prediction of asthma using biomedical signals, particularly focusing on Electrical Impedance Tomography (EIT), a non-invasive technique for assessing lung aeration. EIT provides insights into regional abnormalities and their response to Positive End-Expiratory Pressure (PEEP). Leveraging a PEEP Expiratory Occlusion dataset, we employed machine learning and deep learning techniques to predict disease outcomes. Under ML, Random Forest (RF) model was used to evaluate features importance, while a Feedforward Neural Network (FNN) was utilized to capture complex nonlinear patterns of the data in DL.The results demonstrate that FNN excels in identifying intricate, overlapping patterns, whereas the RF model highlights the significance of individual features. Together, these models achieved high predictive accuracy, offering a promising approach to enhancing the diagnostic process for asthma and COPD. This study highlights the transformative potential of artificial intelligence in disease prediction and rapid diagnostics, paving the way for improved healthcare outcomes.