COPD phenotypes and machine learning cluster analysis: A systematic review and future research agenda

Proteomic Blood Profiles Obtained by Totally Blind Biological Clustering in Stable and Exacerbated COPD Patients

Although Chronic Obstructive Pulmonary Disease (COPD) is highly prevalent, it is often underdiagnosed. One of the main characteristics of this heterogeneous disease is the presence of periods of acute clinical impairment (exacerbations). Obtaining blood biomarkers for either COPD as a chronic entity or its exacerbations (AECOPD) will be particularly useful for the clinical management of patients. However, most of the earlier studies have been characterized by potential biases derived from pre-existing hypotheses in one or more of their analysis steps: some studies have only targeted molecules already suggested by pre-existing knowledge, and others had initially carried out a blind search but later compared the detected biomarkers among well-predefined clinical groups. We hypothesized that a clinically blind cluster analysis on the results of a non-hypothesis-driven wide proteomic search would determine an unbiased grouping of patients, potentially reflecting their endotypes and/or clinical characteristics. To check this hypothesis, we included the plasma samples from 24 clinically stable COPD patients, 10 additional patients with AECOPD, and 10 healthy controls. The samples were analyzed through label-free liquid chromatography/tandem mass spectrometry. Subsequently, the Scikit-learn machine learning module and K-means were used for clustering the individuals based solely on their proteomic profiles. The obtained clusters were confronted with clinical groups only at the end of the entire procedure. Although our clusters were unable to differentiate stable COPD patients from healthy individuals, they segregated those patients with AECOPD from the patients in stable conditions (sensitivity 80%, specificity 79%, and global accuracy, 79.4%). Moreover, the proteins involved in the blind grouping process to identify AECOPD were associated with five biological processes: inflammation, humoral immune response, blood coagulation, modulation of lipid metabolism, and complement system pathways. Even though the present results merit an external validation, our results suggest that the present blinded approach may be useful to segregate AECOPD from stability in both the clinical setting and trials, favoring more personalized medicine and clinical research.

Demystification of artificial intelligence for respiratory clinicians managing patients with obstructive lung diseases

INTRODUCTION

Asthma and chronic obstructive pulmonary disease (COPD) are leading causes of morbidity and mortality worldwide. Despite all available diagnostics and treatments, these conditions pose a significant individual, economic and social burden. Artificial intelligence (AI) promises to support clinical decision-making processes by optimizing diagnosis and treatment strategies of these heterogeneous and complex chronic respiratory diseases. Its capabilities extend to predicting exacerbation risk, disease progression and mortality, providing healthcare professionals with valuable insights for more effective care. Nevertheless, the knowledge gap between respiratory clinicians and data scientists remains a major constraint for wide application of AI and may hinder future progress. This narrative review aims to bridge this gap and encourage AI deployment by explaining its methodology and added value in asthma and COPD diagnosis and treatment.

AREAS COVERED

This review offers an overview of the fundamental concepts of AI and machine learning, outlines the key steps in building a model, provides examples of their applicability in asthma and COPD care, and discusses barriers to their implementation.

EXPERT OPINION

Machine learning can advance our understanding of asthma and COPD, enabling personalized therapy and better outcomes. Further research and validation are needed to ensure the development of clinically meaningful and generalizable models.

Telehealth nursing interventions for phenotypes of older adults with COPD: an exploratory study

Introduction

Inconclusive results exist around the effectiveness of telemonitoring for patients with COPD, and studies recommended conducting subgroup analyses to identify patient phenotypes that could benefit from these services. This exploratory study investigated what type of COPD patients were receiving which type of telenursing interventions more frequently using the telemonitoring platform.

Methods

A sample of 36 older adults with COPD were receiving telenursing services for 12 months and were asked to answer five COPD-symptom related questions and submit their vital signs daily.

Results

Findings revealed two phenotypes of older adults for whom the frequency of telenursing calls and related interventions differed. Although no statistically significant differences were observed in participants' GOLD grades and hospitalizations, cluster one participants used their COPD action plan significantly more frequently, and were in frequent contact with the telenurse.

Discussion

It is paramount that further research is needed on the development of patient phenotypes who may benefit from telemonitoring.

Cluster Analyses From the Real-World NOVELTY Study: Six Clusters Across the Asthma-COPD Spectrum

BACKGROUND

Asthma and chronic obstructive pulmonary disease (COPD) are complex diseases, the definitions of which overlap.

OBJECTIVE

To investigate clustering of clinical/physiological features and readily available biomarkers in patients with physician-assigned diagnoses of asthma and/or COPD in the NOVEL observational longiTudinal studY (NOVELTY; NCT02760329).

METHODS

Two approaches were taken to variable selection using baseline data: approach A was data-driven, hypothesis-free and used the Pearson dissimilarity matrix; approach B used an unsupervised Random Forest guided by clinical input. Cluster analyses were conducted across 100 random resamples using partitioning around medoids, followed by consensus clustering.

RESULTS

Approach A included 3796 individuals (mean age, 59.5 years; 54% female); approach B included 2934 patients (mean age, 60.7 years; 53% female). Each identified 6 mathematically stable clusters, which had overlapping characteristics. Overall, 67% to 75% of patients with asthma were in 3 clusters, and approximately 90% of patients with COPD were in 3 clusters. Although traditional features such as allergies and current/ex-smoking (respectively) were higher in these clusters, there were differences between clusters and approaches in features such as sex, ethnicity, breathlessness, frequent productive cough, and blood cell counts. The strongest predictors of the approach A cluster membership were age, weight, childhood onset, prebronchodilator FEV1, duration of dust/fume exposure, and number of daily medications.

CONCLUSIONS

Cluster analyses in patients from NOVELTY with asthma and/or COPD yielded identifiable clusters, with several discriminatory features that differed from conventional diagnostic characteristics. The overlap between clusters suggests that they do not reflect discrete underlying mechanisms and points to the need for identification of molecular endotypes and potential treatment targets across asthma and/or COPD.

Deep learning diagnostic and severity-stratification for interstitial lung diseases and chronic obstructive pulmonary disease in digital lung auscultations and ultrasonography: clinical protocol for an observational case-control study

BACKGROUND

Interstitial lung diseases (ILD), such as idiopathic pulmonary fibrosis (IPF) and non-specific interstitial pneumonia (NSIP), and chronic obstructive pulmonary disease (COPD) are severe, progressive pulmonary disorders with a poor prognosis. Prompt and accurate diagnosis is important to enable patients to receive appropriate care at the earliest possible stage to delay disease progression and prolong survival. Artificial intelligence-assisted lung auscultation and ultrasound (LUS) could constitute an alternative to conventional, subjective, operator-related methods for the accurate and earlier diagnosis of these diseases. This protocol describes the standardised collection of digitally-acquired lung sounds and LUS images of adult outpatients with IPF, NSIP or COPD and a deep learning diagnostic and severity-stratification approach.

METHODS

A total of 120 consecutive patients (≥ 18 years) meeting international criteria for IPF, NSIP or COPD and 40 age-matched controls will be recruited in a Swiss pulmonology outpatient clinic, starting from August 2022. At inclusion, demographic and clinical data will be collected. Lung auscultation will be recorded with a digital stethoscope at 10 thoracic sites in each patient and LUS images using a standard point-of-care device will be acquired at the same sites. A deep learning algorithm (DeepBreath) using convolutional neural networks, long short-term memory models, and transformer architectures will be trained on these audio recordings and LUS images to derive an automated diagnostic tool. The primary outcome is the diagnosis of ILD versus control subjects or COPD. Secondary outcomes are the clinical, functional and radiological characteristics of IPF, NSIP and COPD diagnosis. Quality of life will be measured with dedicated questionnaires. Based on previous work to distinguish normal and pathological lung sounds, we estimate to achieve convergence with an area under the receiver operating characteristic curve of > 80% using 40 patients in each category, yielding a sample size calculation of 80 ILD (40 IPF, 40 NSIP), 40 COPD, and 40 controls.

DISCUSSION

This approach has a broad potential to better guide care management by exploring the synergistic value of several point-of-care-tests for the automated detection and differential diagnosis of ILD and COPD and to estimate severity. Trial registration Registration: August 8, 2022.

CLINICALTRIALS

gov Identifier: NCT05318599.

EPHX1 and GSTP1 polymorphisms are associated with COPD risk: a systematic review and meta-analysis

Background: Chronic obstructive pulmonary disease (COPD) affects approximately 400 million people worldwide and is associated with high mortality and morbidity. The effect of EPHX1 and GSTP1 gene polymorphisms on COPD risk has not been fully characterized. Objective: To investigate the association of EPHX1 and GSTP1 gene polymorphisms with COPD risk. Methods: A systematic search was conducted on 9 databases to identify studies published in English and Chinese. The analysis was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses reporting guidelines (PRISMA). The pooled OR and 95% CI were calculated to evaluate the association of EPHX1 and GSTP1 gene polymorphisms with COPD risk. The I² test, Q test, Egger's test, and Begg's test were conducted to determine the level of heterogeneity and publication bias of the included studies. Results: In total, 857 articles were retrieved, among which 59 met the inclusion criteria. The EPHX1 rs1051740 polymorphism (homozygote, heterozygote, dominant, recessives, and allele model) was significantly associated with high risk of COPD risk. Subgroup analysis revealed that the EPHX1 rs1051740 polymorphism was significantly associated with COPD risk among Asians (homozygote, heterozygote, dominant, and allele model) and Caucasians (homozygote, dominant, recessives, and allele model). The EPHX1 rs2234922 polymorphism (heterozygote, dominant, and allele model) was significantly associated with a low risk of COPD. Subgroup analysis showed that the EPHX1 rs2234922 polymorphism (heterozygote, dominant, and allele model) was significantly associated with COPD risk among Asians. The GSTP1 rs1695 polymorphism (homozygote and recessives model) was significantly associated with COPD risk. Subgroup analysis showed that the GSTP1 rs1695 polymorphism (homozygote and recessives model) was significantly associated with COPD risk among Caucasians. The GSTP1 rs1138272 polymorphism (heterozygote and dominant model) was significantly associated with COPD risk. Subgroup analysis suggested that the GSTP1 rs1138272 polymorphism (heterozygote, dominant, and allele model) was significantly associated with COPD risk among Caucasians. Conclusion: The C allele in EPHX1 rs1051740 among Asians and the CC genotype among Caucasians may be risk factors for COPD. However, the GA genotype in EPHX1 rs2234922 may be a protective factor against COPD in Asians. The GG genotype in GSTP1 rs1695 and the TC genotype in GSTP1 rs1138272 may be risk factors for COPD, especially among Caucasians.

Severity of Lung Function Impairment Drives Transcriptional Phenotypes of COPD and Relates to Immune and Metabolic Processes

Purpose

This study sought to characterize transcriptional phenotypes of COPD through unsupervised clustering of sputum gene expression profiles, and further investigate mechanisms underlying the characteristics of these clusters.

Patients and methods

Induced sputum samples were collected from patients with stable COPD (n = 72) and healthy controls (n = 15). Induced sputum was collected for inflammatory cell counts, and RNA extracted. Transcriptional profiles were generated (Illumina Humanref-8 V2) and analyzed by GeneSpring GX14.9.1. Unsupervised hierarchical clustering and differential gene expression analysis were performed, and gene alterations validated in the ECLIPSE dataset (GSE22148).

Results

We identified 2 main clusters (Cluster 1 [n = 35] and Cluster 2 [n = 37]), which further divided into 4 sub-clusters (Sub-clusters 1.1 [n = 14], 1.2 [n = 21], 2.1 [n = 20] and 2.2 [n = 17]). Compared with Cluster 1, Cluster 2 was associated with significantly lower lung function (p = 0.014), more severe disease (p = 0.009) and breathlessness (p = 0.035), and increased sputum neutrophils (p = 0.031). Sub-cluster 1.1 had significantly higher proportion of people with comorbid cardiovascular disease compared to the other 3 sub-clusters (92.5% vs 57.1%, 50% and 52.9%, p < 0.013). Through supervised analysis we determined that degree of airflow limitation (GOLD stage) was the predominant factor driving gene expression differences in our transcriptional clusters. There were 452 genes (adjusted p < 0.05 and ≥2 fold) altered in GOLD stage 3 and 4 versus 1 and 2, of which 281 (62%) were also found to be significantly expressed between these GOLD stages in the ECLIPSE data set (GSE22148). Differentially expressed genes were largely downregulated in GOLD stages 3 and 4 and connected in 5 networks relating to lipoprotein and cholesterol metabolism; metabolic processes in oxidation/reduction and mitochondrial function; antigen processing and presentation; regulation of complement activation and innate immune responses; and immune and metabolic processes.

Conclusion

Severity of lung function drives 2 distinct transcriptional phenotypes of COPD and relates to immune and metabolic processes.

Artificial intelligence and its impact on the domains of universal health coverage, health emergencies and health promotion: An overview of systematic reviews

•

An overview of systematic reviews on the application of AI including 129 studies.

•

AI use is prominent in Universal Health Coverage, featuring image analysis in neoplasms.

•

Half of the reviews did not evaluate validation procedures nor reporting guidelines.

•

Risk of bias was only included un a third of the reviews.

•

There is not sufficient evidence to transfer AI to actual healthcare delivery.

Background

Artificial intelligence is fueling a new revolution in medicine and in the healthcare sector. Despite the growing evidence on the benefits of artificial intelligence there are several aspects that limit the measure of its impact in people’s health. It is necessary to assess the current status on the application of AI towards the improvement of people’s health in the domains defined by WHO’s Thirteenth General Programme of Work (GPW13) and the European Programme of Work (EPW), to inform about trends, gaps, opportunities, and challenges.

Objective

To perform a systematic overview of systematic reviews on the application of artificial intelligence in the people’s health domains as defined in the GPW13 and provide a comprehensive and updated map on the application specialties of artificial intelligence in terms of methodologies, algorithms, data sources, outcomes, predictors, performance, and methodological quality.

Methods

A systematic search in MEDLINE, EMBASE, Cochrane and IEEEXplore was conducted between January 2015 and June 2021 to collect systematic reviews using a combination of keywords related to the domains of universal health coverage, health emergencies protection, and better health and wellbeing as defined by the WHO’s PGW13 and EPW. Eligibility criteria was based on methodological quality and the inclusion of practical implementation of artificial intelligence. Records were classified and labeled using ICD-11 categories into the domains of the GPW13. Descriptors related to the area of implementation, type of modeling, data entities, outcomes and implementation on care delivery were extracted using a structured form and methodological aspects of the included reviews studies was assessed using the AMSTAR checklist.

Results

The search strategy resulted in the screening of 815 systematic reviews from which 203 were assessed for eligibility and 129 were included in the review. The most predominant domain for artificial intelligence applications was Universal Health Coverage (N = 98) followed by Health Emergencies (N = 16) and Better Health and Wellbeing (N = 15). Neoplasms area on Universal Health Coverage was the disease area featuring most of the applications (21.7 %, N = 28). The reviews featured analytics primarily over both public and private data sources (67.44 %, N = 87). The most used type of data was medical imaging (31.8 %, N = 41) and predictors based on regions of interest and clinical data. The most prominent subdomain of Artificial Intelligence was Machine Learning (43.4 %, N = 56), in which Support Vector Machine method was predominant (20.9 %, N = 27). Regarding the purpose, the application of Artificial Intelligence I is focused on the prediction of the diseases (36.4 %, N = 47). With respect to the validation, more than a half of the reviews (54.3 %, N = 70) did not report a validation procedure and, whenever available, the main performance indicator was the accuracy (28.7 %, N = 37). According to the methodological quality assessment, a third of the reviews (34.9 %, N = 45) implemented methods for analysis the risk of bias and the overall AMSTAR score below was 5 (4.01 ± 1.93) on all the included systematic reviews.

Conclusion

Artificial intelligence is being used for disease modelling, diagnose, classification and prediction in the three domains of GPW13. However, the evidence is often limited to laboratory and the level of adoption is largely unbalanced between ICD-11 categoriesand diseases. Data availability is a determinant factor on the developmental stage of artificial intelligence applications. Most of the reviewed studies show a poor methodological quality and are at high risk of bias, which limits the reproducibility of the results and the reliability of translating these applications to real clinical scenarios. The analyzed papers show results only in laboratory and testing scenarios and not in clinical trials nor case studies, limiting the supporting evidence to transfer artificial intelligence to actual care delivery.

Hierarchical cluster analysis based on disease-associated manifestations of patients with lymphangioleiomyomatosis: An analysis of the national database of designated intractable diseases of Japan

BACKGROUND

Lymphangioleiomyomatosis (LAM) is a rare multisystem disease with variable manifestations and differing rates of progression among individuals. Classification of its phenotypes is an issue for consideration. We hypothesized that clinical manifestations associated with LAM cluster together and identifying these associations would be useful for identifying phenotypes.

METHODS

Using cross-sectional data from the National Database of Designated Intractable Diseases of Japan, we performed a hierarchical cluster analysis based on disease-associated manifestations.

RESULTS

Four clusters were identified from 404 patients (50.4% of 801 LAM patients registered in 2016). Patients in cluster 1 had only dyspnea on exertion, relatively low lung function, the earliest onset age, and the lowest prevalence of tuberous sclerosis complex (TSC). Those in cluster 2 had various manifestations with the highest prevalence of TSC. Patients in cluster 3 had major respiratory symptoms (cough, sputum, or dyspnea on exertion) or fatigue and the lowest lung function. Those in cluster 4 were asymptomatic and had the latest onset age, shortest disease duration, and relatively high prevalence of TSC. Patients in cluster 1 had the highest rate of receiving mechanistic target of rapamycin (mTOR) inhibitor treatment, suggesting that cluster 1 included those with declining lung function for which mTOR inhibitor treatment was required.

CONCLUSIONS

Hierarchical cluster analysis based on manifestations data identified four clusters. The characteristics of cluster 1 are noteworthy in relation to the indication for mTOR inhibitor treatment. A cluster analysis of accumulated and longitudinal data that allows valid clustering and outcome comparisons is required in the future.

COPD profiles and treatable traits using minimal resources: identification, decision tree and stability over time

Background and objective

Profiles of people with chronic obstructive pulmonary disease (COPD) often do not describe treatable traits, lack validation and/or their stability over time is unknown. We aimed to identify COPD profiles and their treatable traits based on simple and meaningful measures; to develop and validate a decision tree and to explore profile stability over time.

Methods

An observational, prospective study was conducted. Clinical characteristics, lung function, symptoms, impact of the disease (COPD Assessment Test—CAT), health-related quality of life, physical activity, lower-limb muscle strength and functional status were collected cross-sectionally and a subsample was followed-up monthly over six months. A principal component analysis and a clustering procedure with k-medoids were applied to identify profiles. A decision tree was developed and validated cross-sectionally. Stability was explored over time with the ratio between the number of timepoints that a participant was classified in the same profile and the total number of timepoints (i.e., 6).

Results

352 people with COPD (67.4 ± 9.9 years; 78.1% male; FEV1 = 56.2 ± 20.6% predicted) participated and 90 (67.6 ± 8.9 years; 85.6% male; FEV1 = 52.1 ± 19.9% predicted) were followed-up. Four profiles were identified with distinct treatable traits. The decision tree included CAT (< 18 or ≥ 18 points); age (< 65 or ≥ 65 years) and FEV1 (< 48 or ≥ 48% predicted) and had an agreement of 71.7% (Cohen’s Kappa = 0.62, p < 0.001) with the actual profiles. 48.9% of participants remained in the same profile whilst 51.1% moved between two (47.8%) or three (3.3%) profiles over time. Overall stability was 86.8 ± 15%.

Conclusion

Four profiles and treatable traits were identified with simple and meaningful measures possibly available in low-resource settings. A decision tree with three commonly used variables in the routine assessment of people with COPD is now available for quick allocation to the identified profiles in clinical practice. Profiles and treatable traits may change over time in people with COPD hence, regular assessments to deliver goal-targeted personalised treatments are needed.

Digital healthcare in COPD management: a narrative review on the advantages, pitfalls, and need for further research

Chronic obstructive pulmonary disease (COPD) remains a leading cause of morbidity and mortality despite current treatment strategies which focus on smoking cessation, pulmonary rehabilitation, and symptomatic relief. A focus of COPD care is to encourage self-management, particularly during COVID-19, where much face-to-face care has been reduced or ceased. Digital health solutions may offer affordable and scalable solutions to support COPD patient education and self-management, such solutions could improve clinical outcomes and expand service reach for limited additional cost. However, optimal ways to deliver digital medicine are still in development, and there are a number of important considerations for clinicians, commissioners, and patients to ensure successful implementation of digitally augmented care. In this narrative review, we discuss advantages, pitfalls, and future prospects of digital healthcare, which offer a variety of tools including self-management plans, education videos, inhaler training videos, feedback to patients and healthcare professionals (HCPs), exacerbation monitoring, and pulmonary rehabilitation. We discuss the key issues with sustaining patient and HCP engagement and limiting attrition of use, interoperability with devices, integration into healthcare systems, and ensuring inclusivity and accessibility. We explore the essential areas of research beyond determining safety and efficacy to understand the acceptability of digital healthcare solutions to patients, clinicians, and healthcare systems, and hence ways to improve this and sustain engagement. Finally, we explore the regulatory challenges to ensure quality and engagement and effective integration into current healthcare systems and care pathways, while maintaining patients’ autonomy and privacy. Understanding and addressing these issues and successful incorporation of an acceptable, simple, scalable, affordable, and future-proof digital solution into healthcare systems could help remodel global chronic disease management and fractured healthcare systems to provide best patient care and optimisation of healthcare resources to meet the global burden and unmet clinical need of COPD.

Stratification of COPD patients towards personalized medicine: reproduction and formation of clusters

BACKGROUND

The global initiative for chronic obstructive lung disease (GOLD) 2020 emphasizes that there is only a weak correlation between FEV₁, symptoms and impairment of the health status of patients with chronic obstructive pulmonary disease (COPD). Various studies aimed to identify COPD phenotypes by cluster analyses, but behavioral aspects besides smoking were rarely included.

METHODS

The aims of the study were to investigate whether (i) clustering analyses are in line with the classification into GOLD ABCD groups; (ii) clustering according to Burgel et al. (Eur Respir J. 36(3):531-9, 2010) can be reproduced in a real-world COPD cohort; and (iii) addition of new behavioral variables alters the clustering outcome. Principal component and hierarchical cluster analyses were applied to real-world clinical data of COPD patients newly referred to secondary care (n = 155). We investigated if the obtained clusters paralleled GOLD ABCD subgroups and determined the impact of adding several variables, including quality of life (QOL), fatigue, satisfaction relationship, air trapping, steps per day and activities of daily living, on clustering.

RESULTS

Using the appropriate corresponding variables, we identified clusters that largely reflected the GOLD ABCD groups, but we could not reproduce Burgel's clinical phenotypes. Adding six new variables resulted in the formation of four new clusters that mainly differed from each other in the following parameters: number of steps per day, activities of daily living and QOL.

CONCLUSIONS

We could not reproduce previously identified clinical COPD phenotypes in an independent population of COPD patients. Our findings therefore indicate that COPD phenotypes based on cluster analysis may not be a suitable basis for treatment strategies for individual patients.

Fast decliner phenotype of chronic obstructive pulmonary disease (COPD): applying machine learning for predicting lung function loss

Background

Chronic obstructive pulmonary disease (COPD) is a heterogeneous group of lung conditions challenging to diagnose and treat. Identification of phenotypes of patients with lung function loss may allow early intervention and improve disease management. We characterised patients with the ‘fast decliner’ phenotype, determined its reproducibility and predicted lung function decline after COPD diagnosis.

Methods

A prospective 4 years observational study that applies machine learning tools to identify COPD phenotypes among 13 260 patients from the UK Royal College of General Practitioners and Surveillance Centre database. The phenotypes were identified prior to diagnosis (training data set), and their reproducibility was assessed after COPD diagnosis (validation data set).

Results

Three COPD phenotypes were identified, the most common of which was the ‘fast decliner’—characterised by patients of younger age with the lowest number of COPD exacerbations and better lung function—yet a fast decline in lung function with increasing number of exacerbations. The other two phenotypes were characterised by (a) patients with the highest prevalence of COPD severity and (b) patients of older age, mostly men and the highest prevalence of diabetes, cardiovascular comorbidities and hypertension. These phenotypes were reproduced in the validation data set with 80% accuracy. Gender, COPD severity and exacerbations were the most important risk factors for lung function decline in the most common phenotype.

Conclusions

In this study, three COPD phenotypes were identified prior to patients being diagnosed with COPD. The reproducibility of those phenotypes in a blind data set following COPD diagnosis suggests their generalisability among different populations.

A Self-Activated CNN Approach for Multi-Class Chest-Related COVID-19 Detection

Chest diseases can be dangerous and deadly. They include many chest infections such as pneumonia, asthma, edema, and, lately, COVID-19. COVID-19 has many similar symptoms compared to pneumonia, such as breathing hardness and chest burden. However, it is a challenging task to differentiate COVID-19 from other chest diseases. Several related studies proposed a computer-aided COVID-19 detection system for the single-class COVID-19 detection, which may be misleading due to similar symptoms of other chest diseases. This paper proposes a framework for the detection of 15 types of chest diseases, including the COVID-19 disease, via a chest X-ray modality. Two-way classification is performed in proposed Framework. First, a deep learning-based convolutional neural network (CNN) architecture with a soft-max classifier is proposed. Second, transfer learning is applied using fully-connected layer of proposed CNN that extracted deep features. The deep features are fed to the classical Machine Learning (ML) classification methods. However, the proposed framework improves the accuracy for COVID-19 detection and increases the predictability rates for other chest diseases. The experimental results show that the proposed framework, when compared to other state-of-the-art models for diagnosing COVID-19 and other chest diseases, is more robust, and the results are promising.

The cardiovascular phenotype of Chronic Obstructive Pulmonary Disease (COPD): Applying machine learning to the prediction of cardiovascular comorbidities

BACKGROUND

Chronic Obstructive Pulmonary Disease (COPD) is a heterogeneous group of lung conditions that are challenging to diagnose and treat. As the presence of comorbidities often exacerbates this scenario, the characterization of patients with COPD and cardiovascular comorbidities may allow early intervention and improve disease management and care.

METHODS

We analysed a 4-year observational cohort of 6883 UK patients who were ultimately diagnosed with COPD and at least one cardiovascular comorbidity. The cohort was extracted from the UK Royal College of General Practitioners and Surveillance Centre database. The COPD phenotypes were identified prior to diagnosis and their reproducibility was assessed following COPD diagnosis. We then developed four classifiers for predicting cardiovascular comorbidities.

RESULTS

Three subtypes of the COPD cardiovascular phenotype were identified prior to diagnosis. Phenotype A was characterised by a higher prevalence of severe COPD, emphysema, hypertension. Phenotype B was characterised by a larger male majority, a lower prevalence of hypertension, the highest prevalence of the other cardiovascular comorbidities, and diabetes. Finally, phenotype C was characterised by universal hypertension, a higher prevalence of mild COPD and the low prevalence of COPD exacerbations. These phenotypes were reproduced after diagnosis with 92% accuracy. The random forest model was highly accurate for predicting hypertension while ruling out less prevalent comorbidities.

CONCLUSIONS

This study identified three subtypes of the COPD cardiovascular phenotype that may generalize to other populations. Among the four models tested, the random forest classifier was the most accurate at predicting cardiovascular comorbidities in COPD patients with the cardiovascular phenotype.

Phenotypic Variations of Mild-to-Moderate Obstructive Pulmonary Diseases According to Airway Inflammation and Clinical Features

Purpose

Asthma and chronic obstructive pulmonary disease (COPD) are complex and heterogeneous inflammatory diseases. We sought to investigate distinct disease profiles based on clinical, cellular and molecular data from patients with mild-to-moderate obstructive pulmonary diseases.

Patients and Methods

Patients with mild-to-moderate allergic asthma (n=30) and COPD (n=30) were prospectively recruited. Clinical characteristics and induced sputum were collected. In total, 35 mediators were assessed in induced sputum. Logistic regression analysis was conducted to identify the optimal factors that were able to discriminate between asthma and COPD. Further, the data were explored using hierarchical clustering in order to discover and compare clusters of combined samples of asthma and COPD patients. Clinical parameters, cellular composition, and sputum mediators of asthma and COPD were assessed between and within obtained clusters.

Results

We found five clinical and biochemical variables, namely IL-6, IL-8, CCL4, FEV₁/VC ratio pre-bronchodilator (%), and sputum neutrophils (%) that differentiated asthma and COPD and were suitable for discrimination purposes. A combination of those variables yielded high sensitivity and specificity in the differentiation between asthma and COPD, although only FEV₁/VC ratio pre-bronchodilator (%) proven significant in the combined model. In cluster analysis, two main clusters were identified: cluster 1, asthma predominant with evidence of eosinophilic airway inflammation and low level of Th1 and Th2 cytokines; and cluster 2, COPD predominant with elevated levels of Th1 and Th2 mediators.

Conclusion

The inflammatory profile of sputum samples from patients with stable mild-to-moderate asthma and COPD is not disease specific, varies within the disease and might be similar between these diseases. This study highlights the need for phenotyping the mild-to-moderate stages according to their clinical and molecular features.

Development and Assessment of Prediction Models for the Development of COPD in a Typical Rural Area in Northwest China

Objective

This study aimed to construct and evaluate a clinical predictive model for the development of COPD in northwest China's rural areas.

Methods

A cross-sectional study of a natural population was performed in rural northwest China. After assessing demographic and disease characteristics, a clinical prediction model was developed. First, we used the least absolute shrinkage and selection operator regression model to screen possible factors influencing COPD. Then construct a logistic regression model and draw a nomogram. The discriminability of the model was further evaluated by the calibration diagram, C-index and ROC curve system. Clinical benefit was analyzed using the decision curve. Finally, the 1000 bootstrap resamples and Harrell's C-index was used for internal verification of the nomogram.

Results

Among 3249 patients in the local rural natural population, 394 (12.13%) were diagnosed with COPD. The LASSO regression model was used to find the optimal combination of parameters, and the screened influencing factors included age, gender, barbeque, smoking, passive smoking, energy type, ventilation system and Post-Bronchodilator FEV1. These predictors are used to construct a nomogram. C index is 0.81 (95% confidence interval:0.79-0.83). The combination of the calibration curve and ROC curve indicates that the model has high discriminability. The decision curve shows benefits in clinical practice when the threshold probability is >6% and <58%, respectively. The internal verification results using Harrell's C-Index were 0.80 (95% confidence interval: 0.78-0.83).

Conclusion

Combining information such as age, sex, barbeque, smoking, passive smoking, type of energy, ventilation systems, and Post-Bronchodilator FEV1 can be easily used to predict the risk of COPD in local rural areas.

Artificial Intelligence and Machine Learning in Chronic Airway Diseases: Focus on Asthma and Chronic Obstructive Pulmonary Disease

Chronic airway diseases are characterized by airway inflammation, obstruction, and remodeling and show high prevalence, especially in developing countries. Among them, asthma and chronic obstructive pulmonary disease (COPD) show the highest morbidity and socioeconomic burden worldwide. Although there are extensive guidelines for the prevention, early diagnosis, and rational treatment of these lifelong diseases, their value in precision medicine is very limited. Artificial intelligence (AI) and machine learning (ML) techniques have emerged as effective methods for mining and integrating large-scale, heterogeneous medical data for clinical practice, and several AI and ML methods have recently been applied to asthma and COPD. However, very few methods have significantly contributed to clinical practice. Here, we review four aspects of AI and ML implementation in asthma and COPD to summarize existing knowledge and indicate future steps required for the safe and effective application of AI and ML tools by clinicians.

Development and Validation of a Method to Estimate COPD Severity in Multiple Datasets: A Retrospective Study

Introduction

Outcomes in chronic obstructive pulmonary disease (COPD) such as symptoms, hospitalisations and mortality rise with increasing disease severity. However, the heterogeneity of electronic medical records presents a significant challenge in measuring severity across geographies. We aimed to develop and validate a method to approximate COPD severity using the Global Initiative for Chronic Obstructive Lung Disease (GOLD) 2011 classification scheme, which categorises patients based on forced expiratory volume in 1 s, hospitalisations and the modified Medical Research Council dyspnoea scale or COPD Assessment Test.

Methods

This analysis was part of a comprehensive retrospective study, including patients sourced from the IQVIA Medical Research Data [IMRD; incorporating data from The Health Improvement Network (THIN), a Cegedim database] and the Clinical Practice Research Datalink (CPRD) in the UK, the Disease Analyzer in Germany and the Longitudinal Patient Data in Italy, France and Australia. Patients in the CPRD with the complete set of information required to calculate GOLD 2011 groups were used to develop the method. Ordinal logistic models at COPD diagnosis and at index (first episode of triple therapy) were then used to validate the method to estimate COPD severity, and this was applied to the full study population to estimate GOLD 2011 categories.

Results

Overall, 4579 and 12,539 patients were included in the model at COPD diagnosis and at index, respectively. Models correctly classified 74.4% and 75.9% of patients into severe and non-severe categories at COPD diagnosis and at index, respectively. Age, gender, time between diagnosis and start of triple therapy, healthcare resource use, comorbid conditions and prescriptions were included as covariates.

Conclusion

This study developed and validated a method to approximate disease severity based on GOLD 2011 categories that can potentially be used in patients without all the key parameters needed for this calculation.