Prediction models for pulmonary function during acute exacerbation of chronic obstructive pulmonary disease

Abstract

Objective: The pulmonary function test is an effort-dependent test; however, during acute exacerbation of chronic obstructive pulmonary disease (AECOPD), patients are unable to effectively cooperate due to poor health. The present study aimed to establish prediction models that only require demographic and inflammatory parameters to predict pulmonary function indexes: forced expiratory volume in one second (FEV1) and forced vital capacity (FVC). Approach: The goal was to establish prediction models based on multi-output support vector regression. A total of 143 subjects received a peripheral blood examination and pulmonary function test. The demographic and inflammatory parameters were used as input features, and FEV1 and FVC were used as the target features in prediction models. Three models (mixed model, severe model and nonsevere model) were established with FEV1 < 1 l as the threshold of severe episodes of AECOPD. The values of FEV1 and FVC from the pulmonary function tests were compared with the prediction models to validate the performances of the developed prediction models. Main results: The severe and nonsevere models’ prediction performances were better than that of the mixed model. The mean squared errors were lower than 0.05 l2, and the decision coefficients (R 2) were higher than 0.40. The two-tailed t-test results showed that for both severe and nonsevere models, the absolute percentage errors of FEV1 and FVC were within 10%. Significance: Our study shows the feasibility of predicting the pulmonary function indexes FEV1 and FVC with demographic and inflammatory parameters when the pulmonary function test fails to be implemented, which is beneficial for the treatment of AECOPD.

Prediction of Pulmonary Function Parameters Based on a Combination Algorithm

“…In terms of the number of people in the dataset, we used data from 1007 subjects, second only to the 3567 in [12]. With regard to the R 2 index, this study obtained results greater than 0.85, which exceeds the results in the literature [11,13,14]. In terms of the RMSE metrics, our results also go beyond those in the literature [11,14].…”

“…Miyoshi et al (2020) developed regression equations to estimate forced vital capacity (FVC) and forced expiratory volume in one second (FEV1) [13]. Chen et al developed an FEV1 and FVC prediction model based on multi-output support vector regression [14].…”

Quantitative evaluation of disease severity in connective tissue disease-associated interstitial lung disease by dual-energy computed tomography

“…But the results of PFTs weakly reflect the extent of ILD [ 9 ] and could be confounded by the presence of comorbidities, such as pleural disease [ 10 ], chronic airway diseases, pulmonary hypertension, or anemia [ 11 , 12 ]. Furthermore, given that the measures of PFT heavily rely on the cooperation of patients [ 13 ], PFT is not applicable to some severe patients. HRCT can define different radiological patterns and distribution of ILD abnormalities, which is fundamental to the diagnosis and monitoring of ILDs.…”

Quantitative Evaluation of Disease Severity in Connective Tissue Disease-associated Interstitial Lung Disease by Dual-energy Computed Tomography

“…But the results of PFTs weakly re ect the extent of ILD 7 and could be confounded by the presence of comorbidities, such as pleural disease 8 , chronic airway diseases, pulmonary hypertension, or anemia 9,10 . Furthermore, given that the measures of PFT heavily rely on the cooperation of patients 11 , PFT is not applicable to some severe patients. HRCT can de ne different radiological patterns and distribution of ILD abnormalities, which is fundamental to the diagnosis and monitoring of ILDs.…”