Artificial intelligence-aided diagnosis model for acute respiratory distress syndrome combining clinical data and chest radiographs

Abstract

Objective: The aim of this study was to develop an artificial intelligence-based model to detect the presence of acute respiratory distress syndrome (ARDS) using clinical data and chest X-ray (CXR) data.

Method: The transfer learning method was used to train a convolutional neural network (CNN) model with an external image dataset to extract the image features. Then, the last layer of the model was fine-tuned to determine the probability of ARDS. The clinical data were trained using three machine learning algorithms-eXtreme Gradient Boosting (XGB), random forest (RF), and logistic regression (LR)-to estimate the probability of ARDS. Finally, ensemble-weighted methods were proposed that combined the image model and the clinical data model to estimate the probability of ARDS. An analysis of the importance of clinical features was performed to explore the most important features in detecting ARDS. A gradient-weighted class activation mapping (Grad-CAM) model was used to explain what our CNN sees and understands when making a decision.

Results: The proposed ensemble-weighted methods improved the performances of the ARDS classifiers (XGB + CNN, area under the curve [AUC] = 0.916; RF + CNN, AUC = 0.920; LR + CNN, AUC = 0.920; XGB + RF + LR + CNN, AUC = 0.925). In addition, the ML model using clinical data to present the top 15 important features to identify the risk factors of ARDS.

Conclusion: This study developed combined machine learning models with clinical data and CXR images to detect ARDS. According to the results of the Shapley Additive exPlanations values and the Grad-CAM techniques, an explicable ARDS diagnosis model is suitable for a real-life scenario.

Keywords: Acute respiratory distress syndrome; artificial intelligence; chest X-ray; clinical data; ensemble-weighted model; machine learning.

Figure 1.

Patient selection flowchart.

Figure 2.

Model framework overview.

Figure 3.

Examples of chest X-rays (CXRs) from the original dataset and after histogram equalization preprocessing.

Figure 4.

Original image and five enhanced images using data augmentation.

Figure 5.

Example of input image (a) and extracted lung region image (b) cropped by the segmentation model (c), resulting in the reshaped image (d).

Figure 6.

Workflow of the convolutional neural network (CNN) model with transfer learning.

Figure 7.

Convolutional neural network (CNN) model for training accuracy, validation accuracy, training loss, validation loss.

Figure 8.

Receiver operating characteristic (ROC) curves demonstrating the performance of the machine learning models and convolutional neural network (CNN) models for ARDS classification: (a) three machine models using clinical data; (b) two CNN models. Note. ARDS: acute respiratory distress syndrome; AUC: area under the curve.

Figure 9.

Receiver operating characteristic (ROC) curves demonstrating the performance of two ensemble-weighted models: (a) average probability; (b) maximum probability. Note. XGB: eXtreme Gradient Boosting; CNN: convolutional neural network; AUC: area under the curve.

Figure 10.

Feature importance (a) and summary plot (b) of SHAP values.

Figure 11.

Comparison of acute respiratory distress syndrome (ARDS) classification models based on original data and segmented image from two cases: (a) Color visualization of a false negative on original images. (b) Color visualization of true positive on segmented images.

Figure 12.

Venn diagrams represent the effectiveness of classification by XGBoost and convolutional neural network (CNN) classifier. (a) True Positive, (b) True Negative.