1
|
Bodduluri S, Nakhmani A, Reinhardt JM, Wilson CG, McDonald ML, Rudraraju R, Jaeger BC, Bhakta NR, Castaldi PJ, Sciurba FC, Zhang C, Bangalore PV, Bhatt SP. Deep neural network analyses of spirometry for structural phenotyping of chronic obstructive pulmonary disease. JCI Insight 2020; 5:132781. [PMID: 32554922 PMCID: PMC7406302 DOI: 10.1172/jci.insight.132781] [Citation(s) in RCA: 15] [Impact Index Per Article: 3.8] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/16/2019] [Accepted: 06/03/2020] [Indexed: 01/07/2023] Open
Abstract
BACKGROUNDCurrently recommended traditional spirometry outputs do not reflect the relative contributions of emphysema and airway disease to airflow obstruction. We hypothesized that machine-learning algorithms can be trained on spirometry data to identify these structural phenotypes.METHODSParticipants enrolled in a large multicenter study (COPDGene) were included. The data points from expiratory flow-volume curves were trained using a deep-learning model to predict structural phenotypes of chronic obstructive pulmonary disease (COPD) on CT, and results were compared with traditional spirometry metrics and an optimized random forest classifier. Area under the receiver operating characteristic curve (AUC) and weighted F-score were used to measure the discriminative accuracy of a fully convolutional neural network, random forest, and traditional spirometry metrics to phenotype CT as normal, emphysema-predominant (>5% emphysema), airway-predominant (Pi10 > median), and mixed phenotypes. Similar comparisons were made for the detection of functional small airway disease phenotype (>20% on parametric response mapping).RESULTSAmong 8980 individuals, the neural network was more accurate in discriminating predominant emphysema/airway phenotypes (AUC 0.80, 95%CI 0.79-0.81) compared with traditional measures of spirometry, FEV1/FVC (AUC 0.71, 95%CI 0.69-0.71), FEV1% predicted (AUC 0.70, 95%CI 0.68-0.71), and random forest classifier (AUC 0.78, 95%CI 0.77-0.79). The neural network was also more accurate in discriminating predominant emphysema/small airway phenotypes (AUC 0.91, 95%CI 0.90-0.92) compared with FEV1/FVC (AUC 0.80, 95%CI 0.78-0.82), FEV1% predicted (AUC 0.83, 95%CI 0.80-0.84), and with comparable accuracy with random forest classifier (AUC 0.90, 95%CI 0.88-0.91).CONCLUSIONSStructural phenotypes of COPD can be identified from spirometry using deep-learning and machine-learning approaches, demonstrating their potential to identify individuals for targeted therapies.TRIAL REGISTRATIONClinicalTrials.gov NCT00608764.FUNDINGThis study was supported by NIH grants K23 HL133438 and R21EB027891 and an American Thoracic Foundation 2018 Unrestricted Research Grant. The COPDGene study is supported by NIH grants NHLBI U01 HL089897 and U01 HL089856. The COPDGene study (NCT00608764) is also supported by the COPD Foundation through contributions made to an Industry Advisory Committee comprising AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, Novartis, and Sunovion.
Collapse
Affiliation(s)
- Sandeep Bodduluri
- UAB Lung Imaging Core
- UAB Lung Health Center
- Division of Pulmonary, Allergy and Critical Care Medicine, and
| | - Arie Nakhmani
- Department of Electrical and Computer Engineering, University of Alabama at Birmingham, Birmingham, Alabama, USA
| | - Joseph M. Reinhardt
- Department of Biomedical Engineering, University of Iowa, Iowa City, Iowa, USA
| | - Carla G. Wilson
- Department of Biostatistics and Bioinformatics, National Jewish Health, Denver, Colorado, USA
| | - Merry-Lynn McDonald
- UAB Lung Health Center
- Division of Pulmonary, Allergy and Critical Care Medicine, and
| | | | - Byron C. Jaeger
- Department of Biostatistics, University of Alabama at Birmingham, Birmingham, Alabama, USA
| | - Nirav R. Bhakta
- Division of Pulmonary, Critical Care, Allergy and Sleep Medicine, University California, San Francisco, San Francisco, California, USA
| | - Peter J. Castaldi
- Channing Division of Network Medicine, Brigham and Women’s Hospital, Boston, Massachusetts, USA
| | - Frank C. Sciurba
- Division of Pulmonary, Allergy and Critical Care Medicine, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
| | - Chengcui Zhang
- Department of Computer Science, University of Alabama at Birmingham, Birmingham, Alabama, USA
| | | | - Surya P. Bhatt
- UAB Lung Imaging Core
- UAB Lung Health Center
- Division of Pulmonary, Allergy and Critical Care Medicine, and
| |
Collapse
|
2
|
New Spirometry Indices for Detecting Mild Airflow Obstruction. Sci Rep 2018; 8:17484. [PMID: 30504791 PMCID: PMC6269456 DOI: 10.1038/s41598-018-35930-2] [Citation(s) in RCA: 12] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/28/2018] [Accepted: 11/13/2018] [Indexed: 11/17/2022] Open
Abstract
The diagnosis of chronic obstructive pulmonary disease (COPD) relies on demonstration of airflow obstruction. Traditional spirometric indices miss a number of subjects with respiratory symptoms or structural lung disease on imaging. We hypothesized that utilizing all data points on the expiratory spirometry curves to assess their shape will improve detection of mild airflow obstruction and structural lung disease. We analyzed spirometry data of 8307 participants enrolled in the COPDGene study, and derived metrics of airflow obstruction based on the shape on the volume-time (Parameter D), and flow-volume curves (Transition Point and Transition Distance). We tested associations of these parameters with CT measures of lung disease, respiratory morbidity, and mortality using regression analyses. There were significant correlations between FEV1/FVC with Parameter D (r = −0.83; p < 0.001), Transition Point (r = 0.69; p < 0.001), and Transition Distance (r = 0.50; p < 0.001). All metrics had significant associations with emphysema, small airway disease, dyspnea, and respiratory-quality of life (p < 0.001). The highest quartile for Parameter D was independently associated with all-cause mortality (adjusted HR 3.22,95% CI 2.42–4.27; p < 0.001) but a substantial number of participants in the highest quartile were categorized as GOLD 0 and 1 by traditional criteria (1.8% and 33.7%). Parameter D identified an additional 9.5% of participants with mild or non-recognized disease as abnormal with greater burden of structural lung disease compared with controls. The data points on the flow-volume and volume-time curves can be used to derive indices of airflow obstruction that identify additional subjects with disease who are deemed to be normal by traditional criteria.
Collapse
|
3
|
Monforton C. Weight of the evidence or wait for the evidence? Protecting underground miners from diesel particulate matter. Am J Public Health 2006; 96:271-6. [PMID: 16380560 PMCID: PMC1470492 DOI: 10.2105/ajph.2005.064410] [Citation(s) in RCA: 22] [Impact Index Per Article: 1.2] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Accepted: 09/05/2005] [Indexed: 11/04/2022]
Abstract
A coalition of mine operators has used a variety of tactics to obstruct scientific inquiry and impede public health action designed to protect underground miners from diesel particulate matter. These workers are exposed to the highest level of diesel particulate matter compared with any other occupational group. This case study profiles a decade-long saga of the Methane Awareness Resource Group Diesel Coalition to impede epidemiological studies on diesel exhaust undertaken by the National Institute for Occupational Safety and Health and the National Cancer Institute, and to derail a health standard promulgated by the Mine Safety and Health Administration. The case study highlights the coalition's mastery of legislative, judicial, and executive branch operations and the reaction of policymakers.
Collapse
Affiliation(s)
- Celeste Monforton
- Department of Environmental and Occupational Health, George Washington University, 2100 M Street NW, Suite 203, Washington, DC 20037, USA.
| |
Collapse
|
4
|
Castillejos M, Gold DR, Dockery D, Tosteson T, Baum T, Speizer FE. Effects of ambient ozone on respiratory function and symptoms in Mexico City schoolchildren. THE AMERICAN REVIEW OF RESPIRATORY DISEASE 1992; 145:276-82. [PMID: 1736731 DOI: 10.1164/ajrccm/145.2_pt_1.276] [Citation(s) in RCA: 57] [Impact Index Per Article: 1.8] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 12/28/2022]
Abstract
The effects of ambient ozone (O3) on respiratory function and acute respiratory symptoms were evaluated in 143 7- to 9-yr-old schoolchildren followed longitudinally at 1- to 2-wk intervals over a period of 6 months at three schools in Pedregal, Mexico City. The maximum O3 level exceeded the World Health Organization guideline of 80 ppb and the U.S. standard of 120 ppb in every week. For an increase from lowest to highest in the mean O3 level during the 48 hr before spirometry (53 ppb), logistic regression estimated relative odds of 1.7 for a child reporting cough/phlegm on the day of spirometry. For the full population, the mean O3 level during the hour before spirometry, not adjusted for temperature and humidity, predicted a significant decrement in FVC but not in FEV1 or FEF25-75. In contrast, the mean O3 level during the previous 24-, 48-, and 168-h periods predicted significant decrements in FEV1 and FEF25-75 but not in FVC. Ozone was consistently associated with a greater decrement in lung function for the 15 children with chronic phlegm as compared with the children without chronic cough, chronic phlegm, or wheeze. Ozone in the previous 24-, 48-, and 168-h periods predicted decrements in FEV1 for children of mothers who were current or former smokers, but not for children of mothers who were never smokers. Many of these effects were reduced in multiple regression analyses including temperature and humidity, as temperature and O3 were highly correlated.(ABSTRACT TRUNCATED AT 250 WORDS)
Collapse
Affiliation(s)
- M Castillejos
- Universidad Autonoma Metropolitana-X, Mexico City, Mexico
| | | | | | | | | | | |
Collapse
|