Search

Adriana Maldonado-Franco,1 Luis F Giraldo-Cadavid,2,3 Eduardo Tuta-Quintero,2 Mauricio Cagy,4 Alirio R Bastidas Goyes,2 Daniel A Botero-Rosas2 1School of Engineering, Universidad de La Sabana, Chía, Colombia; 2School of Medicine, Universidad de La Sabana, Chía, Colombia; 3Interventional Pulmonology Service, Fundación Neumológica Colombiana, Bogotá, DC, Colombia; 4Biomedical Engineering Program, Universidade Federal do Rio de Janeiro, Rio de Janeiro, BrasilCorrespondence: Daniel A Botero-Rosas, Universidad de La Sabana, Morphophysiology Department, Km 7, Northern Highway, Chía, Cundinamarca, 140013, Colombia, Email Daniel.botero@unisabana.edu.coBackground: Development of new tools in artificial intelligence has an outstanding performance in the recognition of multidimensional patterns, which is why they have proven to be useful in the diagnosis of Chronic Obstructive Pulmonary Disease (COPD).Methods: This was an observational analytical single-centre study in patients with spirometry performed in outpatient medical care. The segment that goes from the peak expiratory flow to the forced vital capacity was modelled with quadratic polynomials, the coefficients obtained were used to train and test neural networks in the task of classifying patients with COPD.Results: A total of 695 patient records were included in the analysis. The COPD group was significantly older than the No COPD group. The pre-bronchodilator (Pre BD) and post-bronchodilator (Post BD) spirometric curves were modelled with a quadratic polynomial, and the coefficients obtained were used to feed three neural networks (Pre BD, Post BD and all coefficients). The best neural network was the one that used the post-bronchodilator coefficients, which has an input layer of 3 neurons and three hidden layers with sigmoid activation function and two neurons in the output layer with softmax activation function. This system had an accuracy of 92.9% accuracy, a sensitivity of 88.2% and a specificity of 94.3% when assessed using expert judgment as the reference test. It also showed better performance than the current gold standard, especially in specificity and negative predictive value.Conclusion: Artificial Neural Networks fed with coefficients obtained from quadratic and cubic polynomials have interesting potential of emulating the clinical diagnostic process and can become an important aid in primary care to help diagnose COPD in an early stage.Keywords: artificial neural networks, machine learning, diagnosis, accuracy, COPD

Ángela María Giraldo-Montoya,1,2,&ast; Carlos A Torres-Duque,1,3,&ast; Luis F Giraldo-Cadavid,4,5 Maria E Laucho-Contreras,1 Angélica González-Flórez,1 Ana María Santos,3 Eduardo A Tuta-Quintero,1,5 Bartolomé R Celli,6 Mauricio González-García1,7 1CINEUMO, Research Center, Fundación Neumológica Colombiana, Bogotá, Colombia; 2School of Medicine, Universidad Tecnológica de Pereira, Pereira, Colombia; 3Biosciences Doctoral, Universidad de La Sabana, Chía, Colombia; 4Medical Department, Fundación Neumológica Colombiana, Bogotá, Colombia; 5Epidemiology and Biostatistics Department, Universidad de La Sabana, Chía, Colombia; 6Harvard Medical School, Boston, MA, USA; 7School of Medicine, Universidad de La Sabana, Chía, Colombia&ast;These authors contributed equally to this workCorrespondence: Carlos A Torres-Duque, Research Department, Fundación Neumológica Colombiana, Bogotá, Colombia, Biosciences Doctoral, Universidad de La Sabana, Chía, Colombia, Email ctorres@neumologica.orgIntroduction: There is a need to better understand the etiotypes of chronic obstructive pulmonary disease (COPD) beyond the tobacco-smoke (TS-COPD). Wood smoke COPD (WS-COPD) is characterized by greater airway compromise, milder emphysema, and slower rate of lung function decline than TS-COPD. However, it is unclear if these two etiotypes of COPD have differences in sputum biomarker concentrations. Objective was to compare sputum levels of selected sputum biomarkers between WS-COPD and TS-COPD, and healthy controls.Methods: Eighty-eight women (69± 12 years) were recruited and classified into: WS-COPD (n=31), TS-COPD (n=29) and controls (n=28). Using ELISA, we determined induced sputum levels of metalloproteinase 9 (MMP-9), chemokine ligand 5 (CCL5), interleukin-8 (IL-8), chemokine ligand 16 (CCL16/HCC-4) and vascular endothelial growth factor (VEGF-1). Differences were analyzed by Kruskal–Wallis and Mann–Whitney-U tests and correlation between airflow limitation and biomarkers by Spearman’s test.Results: At similar degree of airflow obstruction, anthropometrics and medications use, the level of sputum CCL5 was higher in TS-COPD than WS-COPD (p=0.03) without differences in MMP-9, IL-8, CCL16/HCC-4, and VEGF-1. Women with WS-COPD and TS-COPD showed significantly higher sputum levels of MMP-9, IL-8 and CCL5 compared with controls (p< 0.001). FEV1% predicted correlated negatively with levels of MMP-9 (rho:-0.26; P=0.016), CCL5 (rho:-0.37; P=0.001), IL-8 (rho:-0.42; P< 0.001) and VEGF (rho:-0.22; P=0.04).Conclusion: While sputum concentrations of MMP-9, IL-8, and CCL5 were higher in COPD women compared with controls, women with TS-COPD had higher levels of CCL5 compared with those with WS-COPD. Whether this finding relates to differences in pathobiological pathways remains to be determined.Keywords: COPD, wood smoke, biomass fuels, tobacco smoke, biomarkers, sputum, cytokines