AI apply in diagnosing COPD

This week, Dr. Ho, Shu Chuan will introduce the application of Artificial Intelligence for home-based pulmonary rehabilitation in COPD (Chronic Obstructive Pulmonary Disease).

First, she will explain the characteristic of COPD disease. COPD is a complex pathology condition and heterogeneous disease. It is a common, progressive, preventable, and treatable disease that can be categorized by persistent airflow limitation. Chronic airflow limitation is caused by a mixture of small airway disease and parenchymal destruction, and the relative contributions of vary from person to person.

So how does AI apply to diagnosing COPD? AI can diagnose differences in objective lung diseases in Pulmonary Function Tests (PFT) by machine learning. Also, deep learning models can recognize different obstructive patterns in computed tomography (CT). Convolutional Neural networks (CNN) and machine learning have many other kinds of applications such as forced oscillation tests, lung sound analysis, and telemedicine with promising results. Medical imaging plays a key role in evaluating in monitoring lung disease. It was used to diagnose diseases such as COPD and lung cancer. For now, the application of medical images on radiomics such as CT, MRI, and PET has become more mature.

Another example is that deep learning techniques are enabling the COPD phenotyping of complex tree structures like vasculature with an unprecedented solution. For example, research from 2020 used a Convolutional Neural Network (CNN) to diagnose and prognosticate COPD outcomes, including COPD status, Acute Respiratory Disease (ARD) events, and mortalities. The slide shows some differences between COPD and COVID-19 including clinical futures, abnormal investigations, and the possible interventions at different stages of the diseases.

Finally, Do. Ho will introduce the practice of pulmonary rehabilitation. Pulmonary rehabilitation is a comprehensive process, in which patients seek medical health due to different reasons. The treatment plan is usually carefully crafted by the team through the initial assessment and the conditions presented by individual cases. And most programs involved two to three hours of education and exercise, three times a week for at least six weeks. By doing so, we are hoping to decrease the symptoms and increase the quality of life for those who suffer from chronic pulmonary disease.

Share and Learn

• How can AI be applied to diagnose COPD and formulate a pulmonary rehabilitation process through certain characteristics?
• Deep learning models can identify different blocking patterns in computed tomography (CT). What other applications of deep learning models in medicine?