Classification of Exacerbation Frequency in the COPDGene Cohort Using Deep Learning With Deep Belief Networks

Abstract

This study aims to develop an automatic classifier based on deep learning for exacerbation frequency in patients with chronic obstructive pulmonary disease (COPD). A three-layer deep belief network (DBN) with two hidden layers and one visible layer was employed to develop classification models and the models’ robustness to exacerbation was analyzed. Subjects from the COPDGene cohort were labeled with exacerbation frequency, defined as the number of exacerbation events per year. 10,300 subjects with 361 features each were included in the analysis. After feature selection and parameter optimization, the proposed classification method achieved an accuracy of 91.99%, using a 10-fold cross validation experiment. The analysis of DBN weights showed that there was a good visual spatial relationship between the underlying critical features of different layers. Our findings show that the most sensitive features obtained from the DBN weights are consistent with the consensus showed by clinical rules and standards for COPD diagnostics. We thus demonstrate that DBN is a competitive tool for exacerbation risk assessment for patients suffering from COPD.

Predicting the hearing outcome in sudden sensorineural hearing loss via machine learning models

“…In the first step, unsupervised pre‐training is conducted to capture abstract features in a layer‐wise manner as restricted Boltzmann machines using unlabeled data. Afterwards, in supervised fine‐tuning, the whole neural network is optimised on classification by retraining with labelled data . Furthermore, DBN framework automatically extracted informative features for hearing outcome prediction using indices from first lay weights.…”

“…DBN can be seen as a highly sophisticated nonlinear feature extractor where each layer of hidden units learns to capture higher order correlations in the original input data. In this study, a three‐layer DBN with two layers of 100‐100 hidden units were constructed as previously described, reducing the dimensionality of the input feature vectors using series layers of multiple restricted Boltzmann machines. To train the DBN model in a generative approach, a baseline system is built by greedy layer‐wise unsupervised training.…”

“…The number of iteration was optimised to 100 as the accuracy curve remains stable once this value was greater than 100. In this study, the DBN model was developed by adopting the MATLAB‐based GitHub DeepLearnToolBox which is downloadable from: https://github.com/rasmusbergpalm/DeepLearnToolbox …”

“…Until recently, a number of machine learning methods, aiming to uncover hidden patterns, unknown correlations and complex trends, and also simplify the overall data preprocessing work for clinical practice and researchers, have emerged as potential alternatives to LR model, for instance, multilayer perceptron (MLP) and support vector machine (SVM). In particular, an advanced deep learning architecture, deep belief network (DBN) is reported to yield even higher competitive performance in extracting complex nonlinear features where each layer of hidden units learns to representative features that capture higher order correlations . Therefore, this study aimed to investigate whether, and in what manner, the machine learning techniques, in particular the deep learning algorithms outperform the traditional LR model for predicting the hearing outcome in SSHL.…”

Machine Learning and Deep Neural Networks in Thoracic and Cardiovascular Imaging

“…Other promising areas of thoracic research include the use of machine learning for aortic segmentation, thrombus detection, and fibrotic and COPD disease classification. [45][46][47]76…”