Analyzing the use of artificial intelligence for the management of chronic obstructive pulmonary disease (COPD)

Highlights

• The development of AI applications to manage COPD has increased considerably.

• Diagnosis, severity classification, and prevention are the main targets.

• Demographics variables, symptoms and spirometric values are frequently used for the diagnosis.

• Neural nets and decision trees were the AI algorithms more implemented.

• The mean and median values of all the performance metrics were between 80% and 90%.

Abstract

Objective

Chronic obstructive pulmonary disease (COPD) is a disease that causes airflow limitation to the lungs and has a high morbidity around the world. The objective of this study was to evaluate how artificial intelligence (AI) is being applied for the management of the disease, analyzing the objectives that are raised, the algorithms that are used and what results they offer.

Methods

We conducted a scoping review following the Arksey and O'Malley (2005) and Levac et al. (2010) guidelines. Two reviewers independently searched, analyzed and extracted data from papers of five databases: Web of Science, PubMed, Scopus, Cinahl and Cochrane. To be included, the studies had to apply some AI techniques for the management of at least one stage of the COPD clinical process. In the event of any discrepancy between both reviewers, the criterion of a third reviewer prevailed.

Results

380 papers were identified through database searches. After applying the exclusion criteria, 67 papers were included in the study. The studies were of a different nature and pursued a wide range of objectives, highlighting mainly those focused on the identification, classification and prevention of the disease. Neural nets, support vector machines and decision trees were the AI algorithms most commonly used. The mean and median values of all the performance metrics evaluated were between 80% and 90%.

Conclusions

The results obtained show a growing interest in the development of medical applications that manage the different phases of the COPD clinical process, especially predictive models. According to the performance shown, these models could be a useful complementary tool in the decision-making by health specialists, although more high-quality ML studies are needed to endorse the findings of this study.

Introduction

Chronic Obstructive Pulmonary Disease (COPD) is a common, preventable, and treatable disease that is characterized by persistent respiratory symptoms and airflow limitation that is due to airway and/or abnormalities usually caused by significant exposure to noxious particles or gases [1]. COPD is used as an umbrella term to refer to a set of lung conditions characterized by progressive respiratory distress, mainly emphysema, chronic bronchitis, and asthma.

According to data from the World Health Organization (WHO), COPD is the third leading cause of death worldwide, causing 3.23 million deaths in 2019. It is estimated that over 80% of these deaths occurred in low and middle-income countries [2]. Overall, the morbidity of COPD is high and increases with age, being higher in men than in women. It can also be affected by other concomitant conditions (e.g., cardiovascular diseases, diabetes mellitus, etc.). COPD could also be the cause of the development of other comorbidities at an early age [3].

There are different risk factors that can influence the development and progression of COPD. These factors are related to individual and environmental factors. In the first group are genetic factors (mainly a severe deficiency of apha-1 antitrypsin) [4], age, sex, and lung development [1]. The second group includes those factors that involve continuous exposure to noxious particles such as sulfur dioxide, organic and inorganic dusts, chemical agents, fumes [5] and cigarette smoking that is the most commonly encountered risk factor for COPD across the world [6]. Symptoms of COPD do not usually appear until significant lung damage occurs, and they usually get worse over time. The most frequent symptoms are dyspnea, cough, sputum production, wheezing, and chest tightness [7], [8], [9]. These symptoms often occur in more intense episodes that can last from days to weeks. These exacerbations require urgent medical attention (even hospitalization), and can sometimes cause death. The appearance of any of these symptoms may suggest the presence of COPD, which needs to be confirmed with a spirometry test. To unify criteria and create a global consensus on the strategy to follow for the diagnosis, treatment, and prevention of COPD some initiatives have emerged, among which the Global Initiative for Chronic Obstructive Lung Disease (GOLD) [1] stands out. It defines four disease stages based on the severity of airflow limitation (mild, moderate, severe, very severe) and a tailored approach for the treatment based on the level of symptoms and risk for exacerbations.

In the last decades, the rise of new technologies has led to the emergence of medical applications for different purposes such as support in diagnosis, severity classification, remote monitoring of chronic patients or prediction of exacerbations. Some of these applications are based on clinical guidelines such as GOLD or GesEPOC [10] for the management of COPD, but others have incorporated artificial intelligence (AI) algorithms for the inference of new conclusions from the information provided by clinical trials or medical databases. Therefore, there is a need to review the existing literature to compile what kind of applications exist for COPD management, focusing mainly on what is the objective they pursue, what AI algorithms they use, and their effectiveness.