Home A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models
Article Open Access

A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models

  • Madhumita Pal , Ranjan K. Mohapatra EMAIL logo , Ashish K. Sarangi , Alok Ranjan Sahu ORCID logo , Snehasish Mishra ORCID logo , Alok Patel , Sushil Kumar Bhoi ORCID logo EMAIL logo , Ashraf Y. Elnaggar , Islam H. El Azab , Mohammed Alissa and Salah M. El-Bahy
Published/Copyright: February 4, 2025
Download Article (PDF)
Open Medicine
From the journal Open Medicine Volume 20 Issue 1

Abstract

Background

The highly infectious coronavirus disease 2019 (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2, the seventh coronavirus. It is the longest pandemic in recorded history worldwide. Many countries are still reporting COVID-19 cases even in the fifth year of its emergence.

Objective

The performance of various machine learning (ML) and deep learning (DL) models was studied for image-based classification of the lungs infected with COVID-19, pneumonia (viral and bacterial), and normal cases from the chest X-rays (CXRs).

Methods

The K-nearest neighbour and logistics regression as the two ML models, and Visual Geometry Group-19, Vision transformer, and ConvMixer as the three DL models were included in the investigation to compare the brevity of the detection and classification of the cases.

Results

Among the investigated models, ConvMixer returned the best result in terms of accuracy, recall, precision, F1-score and area under the curve for both binary as well as multiclass classification. The pre-trained ConvMixer model outperformed the other four models in classifying. As per the performance observations, there was 97.1% accuracy for normal and COVID-19 + pneumonia-infected lungs, 98% accuracy for normal and COVID-19 infected lungs, 82% accuracy for normal + bacterial + viral infected lungs, and 98% accuracy for normal + pneumonia infected lungs. The DL models performed better than the ML models for binary and multiclass classification. The performance of these studied models was tried on other CXR image databases.

Conclusion

The suggested network effectively detected COVID-19 and different types of pneumonia by using CXR imagery. This could help medical sciences for timely and accurate diagnoses of the cases through bioimaging technology and the use of high-end bioinformatics tools.

Keywords: COVID-19; chest X-ray images; ML and DL models; binary and multiclass classification

1 Introduction

Coronavirus disease 2019 (COVID-19) is a highly infectious disease caused by severe acute respiratory syndrome coronavirus (SARS-CoV-2), a coronavirus strain [1,2]. First reported in December 2019 from Wuhan City, China, the scenario became global, for which the World Health Organisation (WHO) officially declared the outbreak as a pandemic in March 2020 [3]. Several COVID-19 waves ensued globally as multiple viral variants emerged, particularly the variants of concern that affected almost every region around the globe [4,5]. A total of 776,696,616 COVID-19 cases with 7,072,509 deaths were reported globally as of 20 October 2024 (https://data.who.int/dashboards/covid19/cases?n=o). Some countries are still reporting COVID-19 infection cases even in the fifth year of the first report of human infection by the virus [6]. The pandemic reshaped the global economy, community health, family life, jobs and employment, education, and social and cultural ethos [7,8]. The impact varied between countries at the global level as this pandemic triggered the largest global economic crisis ever, with increasing poverty and inequalities [9,10]. Further, it presented unprecedented pressure on the health infrastructure that challenged public health and diminished livelihood options, food security, and nutrition provisions, especially in low-income countries with less purchasing capacity [11].

Foolproof diagnostic options are essential tools for swift pandemic response and to activate community health measures. The three widely used tests for diagnosis to detect COVID-19 cases are molecular (PCR), antigen rapid detection, and antibody-based [12]. PCR (polymerase chain reaction) is highly sensitive and specific in detecting RNA viruses. PCR is recommended by the WHO, especially to confirm symptomatic cases. Although antigen-based rapid test detects viral proteins; however, the test is less sensitive compared to molecular tests. The antibody-based test detects the antibody titre in the subject as the response to infection or vaccination and could be a useful public health surveillance tool. These three diagnosis approaches played a crucial role in transitioning from pandemic response to pandemic control. Among these, PCR (especially, reverse transcription polymerase chain reaction; RT-PCR) is the most reliable technique and is widely being employed in diagnosing COVID-19 since the outbreak began [13,14]. Although RT-PCR sensitively and quantitatively detects SARS-CoV-2, it needs skilled clinical laboratory staff with a complicated procedure that costs US$50–100 per test on average, which may not be suitable for low-income countries [15]. Many countries lack adequate supply of RT-PCR test kits, and hence, an alternative automatic diagnostic system for early detection of COVID-19 to prevent its further spread is essential.

Pneumonia is a frequently encountered lung infection, being severe in the elderly and children below 5 years. It hinders a patient’s oxygen intake ability through the lungs into the bloodstream. Pneumonia could be detected by employing numerous biotechniques like chest X-ray (CXR), computer-aided tomography, and magnetic resonance imaging. These need expert radiologists for the purpose, which could often lead to delayed or misdiagnosis. Similar to influenza, clinical investigations have confirmed that COVID-19 affects the lower part of the respiratory tract, especially the lungs [16]. The computed tomography (CT) scan of the chest is an effective imaging technique to diagnose lung-related diseases. However, CXR is a widely accepted alternative in hospitals due to its faster imaging time and low cost compared to the CT scan [16,17]. Thus, CXR images could be very useful in the early diagnosis of COVID-19 and probably could help in prediagnosis and prodiagnosis, too.

Radiologists frequently struggle to distinguish COVID-19 from other pulmonary conditions and community-acquired pneumonia based simply on the X-ray images and CT scans [18]. To promptly confirm COVID-19 cases, the artificial intelligence (AI) approach seems to be appealing to researchers and physicians owing to the expected high accuracy and a lower operational cost [19]. Researchers have employed the open-source COVID-19 database to gather and evaluate radiography data. The digitised X-ray image versions are typically used to classify automatically using machine or deep learning (DL) tools alongside traditional image processing tools [20]. AI technologies are promising in resolving numerous issues with the aid of novel machine learning (ML) and DL tools. It includes enhanced access to high-quality healthcare, especially in rural and low-income areas; resolving the skewed issue of the number of patients and skilled physicians; enhancing the efficiency and training of the healthcare personnel especially engaged in complex procedures; and facilitating the large-scale delivery of personalised healthcare.

Given the scenario, ML- and DL-based computer-aided diagnostic approaches are significant to timely and accurately detect COVID-19 and pneumonia. As the X-ray technique is cheaper compared to other techniques, it was chosen in this investigation as base image data. Pneumonia and COVID-19 detection by using ML or DL techniques have been reported. However, comparative studies on pneumonia and COVID-19 detection using ML and DL simultaneously and comparing the performance are few. This study compared the performance of ML and DL techniques to diagnose COVID-19 and pneumonia using CXR images. Two ML models (the K-nearest neighbour (K-NN) and logistics regression (LR)) and three DL models (Visual Geometry Group-19 (VGG-19), Vision transformer (ViT) and ConvMixer) were chosen to classify pneumonia and COVID-19 CXR images, with images of healthy individuals as the control. The study was conducted with the following as the primary objectives:

  • To implement ML/DL model for binary and multiclass classification of CXR images of normal, COVID-19, and pneumonia cases

  • To differentiate COVID-19-infected lung images from pneumonia and normal CXR images

  • To classify and differentiate pneumonia subtypes from CXR images using ML/DL models

  • To compare the performance of ML and DL models in terms of accuracy, recall, precision, F1-score, and area under the curve (AUC)

  • To compare the performance of the proposed models with other reported models

2 Methodology

The preparatory steps that were carried out for CXR image classification were data collection, image preprocessing to reduce the “noise,” implementation of ML/DL models, and measuring the performance of the models in terms of accuracy, recall, precision, F1-score, and AUC.

2.1 Data collection

Two datasets were chosen for the investigative study. The CXR image dataset was sourced from https://data.mendeley.com/datasets/rscbjbr9sj/2. This dataset contained 5,232 X-ray images of paediatric (1–5 years) cases from Guangzhou Medical Centre, China. It contained X-ray images of the normal chest, the bacterial pneumonia-infected chest, and the viral pneumonia-infected chest. Out of the total 5,856 images in the dataset, 1,583 were normal, and 4,273 were pneumonia cases, of which 2,538 were bacterial pneumonia, and 1,735 were viral pneumonia cases. The second dataset was collected from https://github.com/ieee8023/covid-chestxray-dataset. It consisted of 10,192 normal images, 3,616 COVID-19-positive cases, 6,012 lung opacity (non-COVID lung infection), and 1,345 viral pneumonia images. As the dataset collected from these two public repositories was not balanced, the synthetic minority oversampling technique was used to balance them; 80% of the data was used to train the models, and 20% was used to test the models.

2.2 Image pre-processing

Pre-processing is the first and critical step wherein the images are moderated, and quality is enhanced to improve the performance of computer-assisted diagnosis. Here, the necessary conversion of the raw input images into the correct format is done before feeding to the DL classifiers. This step is very important as the clinical datasets available online to feed the network classifiers often had images of varying size, shape, and contrast. The images ought to be of the same size as the network classifier’s input before feeding. Thus, the images need rescaling or resizing to fit the intended input format.

During the image pre-processing phase, the quality of images was enhanced, and the “noise” was reduced by using contrast-limited adaptive histogram equalisation and numerous other data augmentation techniques like normalisation, resizing, horizontal random flip, random rotation, and zoom. Data augmentation techniques were implemented on the image database to address the over-fitting issue. Normalisation, resizing, horizontal random flip, random rotation with factor 0.2, and zoom (height factor = 0.2, width factor = 0.2) were the steps followed during augmentation.

2.3 Performance parameters employed to compare the ML/DL models

Two ML models (LR and k-NN) and three DL models (VGG19, ViT, and Convmixer) were implemented to detect and classify CXR images. The performance of these models was evaluated in terms of accuracy, recall, precision, F1-score, and AUC employing the following equations. The definition of each performance evaluation matrix is detailed in Table 1:

(1) Precision = True positive True positive + False positive ,

(2) Recall = True positive True positive + False negative ,

(3) F 1-score = 2 × precision × recall precision × recall ,

(4) Accuracy = True positive + True negative True positive + True negative + False positive + False negative .

Table 1

Various training parameters considered to train the models

Model Training parameters
LR Sigmoid function
k-NN Neighbours = 7
ViT Learning rate = 0.001
Weight decay = 0.0001
Batch size = 20
Image size = 72
Patch size = 6
Transformer layer = 8
MLP head units = [2,048, 1,024]
ConvMixer Learning rate = 0.001
Weight decay = 0.0001
Loss = sparse categorical loss function
Optimiser = Adam
Batch size = 8
Epoch = 50
Image size = 150
Filters = 256
Depth = 8
Kernel size = 5
Number of classes = 3
Global average pooling
Activation = softmax
VGG-19 Relu activation function
Batch size = 32
Epoch = 50
Trainable parameters = 20,024,384 (76.39 MB)

The AUC is the curve between the true-positive rate and the false-positive rate. AUC was used to compare the performances of two separate models using a roughly balanced dataset.

2.4 Description of the ML and DL models employed in the study

LR and k-NN as ML models, ViT, VGG-19, and ConvMixer models were used to detect COVID-19 and pneumonia infection from the CXR images. The training parameters used to test the models are detailed in Table 1, and the details of each of the five models are provided in separate sections below.

Total parameters: 20,024,384 (76.39 MB)

Trainable parameters: 20,024,384 (76.39 MB)

Non-trainable parameters: 0 (0.00 Byte)

2.4.1 LR

LR is implemented for the linearly separable data. It is a supervised ML model [21] used for binary classification tasks. It provides discrete outcomes such as 0 and 1. It is used to predict the probability of occurrence of certain events or classes of events. Single independent variables or multiple independent variables are used to predict the outcome of logistic regression. The predicted probability value is converted into 1 or 0 using the sigmoid activation function [22] (Figure 1). The sigmoid function is provided in equation (5):

(5) Ψ ( z ) = 1 1 + e − z .

Figure 1 Output of sigmoid function for the predicted probability value.
Figure 1

Output of sigmoid function for the predicted probability value.

2.4.2 k-NN

k-NN is the most popular and simple distance-based supervised ML algorithm used for regression as well as classification tasks, although commonly used for classification tasks. It is a lazy nonparametric learning model [23] as it learns during the testing phase and stores data only during the training phase. The k-NN model follows the nearest data point and classifies the data point accordingly [21]. Its classification is based on the consensus of its k neighbours. The case assigned to the class shares the most occurrences with its k nearest neighbour, as determined by a distance function. Different distance matrices like Manhattan distance, Euclidean distance, cosine distance, and Minkowski distance are used to ascertain the distance between all the training data and new data points. The k value could be properly chosen to get better classification results during the testing phase, or else the model either overfits (k = 1) or underfits (higher k value).

2.4.3 ViT

An advanced DL architecture created, especially for visual recognition applications, is the ViT Model [24]. It is a novel model combining the capabilities of transformer models with computer vision that was originally developed for natural language processing. The self-attention mechanism at the core of the operational logic of the ViT Model allowed it to recognize contextual information and global interdependencies within an image. ViT Model leverages the attention mechanism [25] to directly extract relevant visual features from unprocessed pictures, unlike conventional convolutional neural networks (CNNs) that depend on hand-crafted features. An input image is divided into smaller patches by the ViT Model, treating them as consecutive tokens. These patches were then fed into a transformer encoder that has several layers of feed-forward and self-attention neural networks (Figure 2). The model can selectively focus on different parts of an image and understand the complex relationships through the self-attention mechanism, and the feed-forward networks process the attained data to produce meaningful visual embeddings. ViT Model learns to use large image datasets (like ImageNet with large labeled images) during the training phase. The model learns to predict the right class labels for the photos through supervised learning, fine-tuning its parameters using gradient descent and back-propagation techniques. In addition, the ViT Model presents notable interpretability and scalability benefits. It is capable of handling a wide range of image sizes and complexity levels with ease. By examining and explaining the decision-making process of the model, researchers and practitioners could better understand the variables influencing the model’s predictions, made possible by the self-attention mechanism.

Figure 2 The structure of the ViT model.
Figure 2

The structure of the ViT model.

In summary, the (Google) ViT Model uses the remarkable combined power of transformer models for visual identification tasks, epitomising a paradigm shift in computer vision. Its capacity to recognise global interdependencies and self-attention mechanisms has led to advancements in various application disciplines. The ViT Model has transformed computer vision by utilising large datasets and cutting-edge training techniques, opening up new avenues for investigations and useful applications.

2.4.4 ConvMixer

ConvMixer bears many similarities to ViT (and MLP-Mixer); it operates directly on patches, preserves an equal-resolution-and-size representation across all layers [25], does not down-sample the representation at successive layers, and differentiates between “spatial mixing” and “channel-wise mixing” of information (Figure 3). ConvMixer performs all these using ordinary convolutions, unlike ViT and MLP-Mixer. ConvMixer consists of a patch-embedding layer and a simple, fully convolutional block applied repeatedly. Convolution with c in input channels, h output channels, p kernel size, and p stride could be used to implement patch embeddings with p patch size and h embedding dimension (equation (6)).

(6) z 0 = BN ( α Conv cin → h ( X , Stride = p , kernel size= p ) .

Figure 3 The architecture of the ConvMixer model.
Figure 3

The architecture of the ConvMixer model.

2.4.5 Visual geometric group-19 (VGG-19)

A VGG network is built using minuscule convolutional filters [26]. Thirteen convolutional layers and six fully connected layers make up the VGG-19 (that is, 19 layers deep). It consists of convolutional, hidden, and fully connected layers. The network structure of the VGG-19 model is shown in Table S1.

Quickly examining the VGG architecture, as the 224 × 224 picture is fed into the VGG Net (very deep convolutional network), the developers chop out the central 224 × 224 patch for each image for the ImageNet competition to maintain a uniform input size of the image. The convolutional layers of VGG make use of a minimal receptive field, i.e. 33, the least size that still captures left-right and up-down. Additionally, a 1X1 convolution filter is used to transform the input linearly. Rectified linear unit activation function and a significant AlexNet innovation that reduces training duration come next. ReLU gives an output if the input is positive, else it gives zero output. The convolution stride is fixed at one pixel to maintain the spatial resolution after convolution (stride is the number of pixel shifts over the input matrix). The VGG Net consists of three fully connected layers, each of which has 4,096 channels, and the third layer has 1,000 channels, one for each class.

2.5 The simulation software

Google Collaboratory, a “Cloud” platform-hosted free Jupyter notebook, was used to conduct the study. Free GPU access and a zero-configuration interface were also included to create and run Python code straight from the browser using well-known Python packages to aid in data analysis. The Colab was used for visualization purposes. The input datasets could be images and train classifiers to assess the classifiers’ performance. Tesla T4 device was used and the computing capability was 7.5.12.7 GB RAM used for the simulation job. Pci bus id: 0000:00:04.0. Cloud server was employed to use the GPU’s power.

  1. Ethical approval: The nature of this article does not require any ethical approval.

3 Results of the simulation work

The values of the performance metrics parameters like accuracy, precision, recall, F1-score, and AUC are provided in Table 2. The AUC curve of each ML and DL model is shown in Figure 4. From the AUC curves (Figure 4), it was demonstrated that ConvMixer gave the highest AUC values for class 0 = 0.97, class 1 = 0.99, and class 2 = 0.97 compared to other ML/DL models for multiclass classification. The output of the multiclass classification of the ConvMixer model is shown in Figure 4. The accuracy obtained by the ConvMixer model was 97.09.

Table 2

ML-DL model’s performance evaluation for multiclass classification of CXRs

ML-DL model Class type* Precision Recall F1-score AUC score Accuracy
Performance evaluation of the two ML models
Logistic regression 0 0.88 0.92 0.90 0.83 85.02
1 0.85 0.87 0.86 0.93
2 0.74 0.66 0.70 0.79
k-nearest neighbour 0 0.85 0.96 0.90 0.80 84.96
1 0.93 0.88 0.90 0.94
2 0.82 0.53 0.64 0.75
Performance evaluation of the three DL models
VGG-19 0 0.96 0.95 0.95 0.93 93.53
1 0.89 0.99 0.94 0.99
2 0.89 0.88 0.89 0.92
ViT 0 0.93 0.97 0.95 0.91 92.61
1 0.95 0.93 0.94 0.96
2 0.89 0.81 0.85 0.89
ConvMixer 0 0.98 0.98 0.98 0.97 97.09
1 0.93 0.99 0.96 0.99
2 0.95 0.95 0.95 0.97

*Class 0: normal (control); Class 1: viral pneumonia; Class 2: COVID-19.

Figure 4 ROC curves: (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer.
Figure 4

ROC curves: (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer.

The ROC curve represented the ability of each model to diagnose at varying threshold levels. It is the curve between the true positive rate (no. of positive samples that are correctly predicted by an ML model) and the false positive rate (no. of actual negative samples that are incorrectly predicted as positive by the ML model). The ability of a model to accurately diagnose depends on the area covered by each ROC curve. More area meant more classification ability of the ML model. As observed in Figure 4, the AUC scores obtained for Class 0, Class 1, and Class 2 by the LR model were, respectively, 0.83, 0.93, and 0.79. The values similarly were 0.80, 0.94, and 0.75, respectively, in k-NN; 0.85, 0.91, and 0.82 in CNN; 0.93, 0.99, and 0.92 in VGG-19; 0.91, 0.96, and 0.89 in ViT; and 0.97, 0.99, and 0.97 in the ConvMixer models.

As seen from the output of the ConvMixer model (Figure 5), a normal CXR (left panel) portrayed clear lungs with no abnormal opacification. Bacterial pneumonia (middle set) exhibited a focal lobar consolidation (in the right upper lobe in this case), whereas viral pneumonia (right panel) set showed a distinct and more diffused “interstitial” pattern in both the lungs. The binary (COVID/normal) classification output of the ConvMixer model is shown in Figure 6.

Figure 5 The normal, COVID, and viral pneumonia outputs from CXRs by ConvMixer.
Figure 5

The normal, COVID, and viral pneumonia outputs from CXRs by ConvMixer.

Figure 6 The output of the ConvMixer model for binary (COVID/normal) classification.
Figure 6

The output of the ConvMixer model for binary (COVID/normal) classification.

The values of the performance parameters like precision, recall, F1-score, AUC, and accuracy in ML-DL models are provided in Table 3. The values obtained against the performance parameters showed that ConvMixer had the maximum accuracy (98%) and 0.97 AUC for COVID-19 infected lungs classification against the normal lungs.

Table 3

Performance comparison of the models for binary classification

ML-DL model Class type* Precision Recall F1-score AUC score Accuracy
Logistic regression 0 0.91 0.91 0.91 0.82 86.89
1 0.75 0.75 0.75
k-NN 0 0.87 0.97 0.91 0.77 86.53
1 0.87 0.57 0.69
VGG-19 0 0.98 0.95 0.96 0.94 94.71
1 0.87 0.94 0.90
ViT 0 0.96 0.98 0.97 0.93 95.87
1 0.94 0.88 0.91
ConvMixer 0 0.99 0.99 0.99 0.97 98.00
1 0.96 0.96 0.96

*Class 0: normal (control); Class 1: viral COVID-19.

The ROC curves of the models to diagnose COVID-19-infected lungs from normal lungs are shown in Figure 7.

Figure 7 ROC curves of (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer differentiating COVID-19-infected lungs from the normal ones.
Figure 7

ROC curves of (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer differentiating COVID-19-infected lungs from the normal ones.

The comparative performance values of the considered two ML and three DL models to diagnose and differentiate the bacterial and viral pneumonia against the control (normal) are detailed in Table 4.

Table 4

Comparative performance of the models to detect bacterial and viral pneumonia

ML-DL model Class type* Precision Recall F1-score AUC score Accuracy
Performance evaluation of the ML models
Logistic regression 0 0.84 0.89 0.87 0.91 74.57
1 0.75 0.81 0.78 0.79
2 0.59 0.47 0.52 0.68
K-NN 0 0.88 0.91 0.90 0.93 76.27
1 0.79 0.80 0.80 0.80
2 0.57 0.53 0.55 0.70
Performance evaluation of the DL models
VGG-19 0 0.94 0.91 0.92 0.94 77.13
1 0.73 0.89 0.80 0.79
2 0.66 0.40 0.50 0.67
ViT 0 0.94 0.94 0.94 0.96 76.62
1 0.75 0.82 0.78 0.79
2 0.59 0.48 0.53 0.68
ConvMixer 0 0.87 0.93 0.90 0.94 81.56
1 0.79 0.89 0.84 0.84
2 0.79 0.55 0.65 0.75

*Class 0: normal (control); Class 1: viral pneumonia; and Class 2: COVID-19.

Figure 8 shows the classifiability of the various types of pneumonia (bacterial and viral) by ConvMixer from the CXR image database. From the output, it was observed that the normal CXR images (left panel) portrayed healthy lungs clearly without any abnormal opacification. The bacterial pneumonia (middle panel) typically exhibited a focal lobar consolidation, in this case in the right upper lobe (white arrows), and the viral pneumonia (right panel) was distinct with a more diffuse “interstitial” pattern in both the lungs.

Figure 8 ConvMixer model outputs classifying viral pneumonia, bacterial pneumonia, and normal lung.
Figure 8

ConvMixer model outputs classifying viral pneumonia, bacterial pneumonia, and normal lung.

Analysing the CXR images, the AUC scores obtained in the LR model for normal (Class 0), bacterial (Class 1), and viral (Class 2) pneumonia lungs were, respectively, 0.91, 0.79, and 0.68 (Figure 9). The figures similarly were 0.93 (Class 0), 0.80 (Class 1), and 0.70 (Class 2) in the k-NN model; 0.92 (Class 0), 0.79 (Class 1), and 0.88 (Class 2) in the CNN model; 0.94 (Class 0), 0.79 (Class 1), and 0.67 (Class 2) in the VGG-19 model; 0.96 (Class 0), 0.79 (Class 1), and 0.68 (Class 2) in the ViT model; and 0.94 (Class 0), 0.84 (Class 1), and 0.75 (Class 2) in the ConvMixer model.

Figure 9 ROC of (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer to detect normal, bacterial-infected, and viral-infected lungs.
Figure 9

ROC of (a) LR, (b) k-NN, (c) VGG-19, (d) ViT, and (e) ConvMixer to detect normal, bacterial-infected, and viral-infected lungs.

The performance comparison of the models for pneumonia-affected lung and normal lung detection is given in Table 5. The maximum AUC and accuracy of 0.97 and 97.61% were obtained by the ConvMixer model for the classification of pneumonia-infected lungs from normal lungs, followed by the VGG-19 model with an AUC and accuracy of 0.95 and 96.75, respectively.

Table 5

Comparing the studied models’ performances to detect pneumonia-infected lungs

ML-DL model Class type* Precision Recall F1-score AUC score Accuracy
Logistic regression 0 0.88 0.76 0.84 0.87 92.32
1 0.92 0.96 0.95
k-NN 0 094 0.86 0.90 0.92 86.53
1 0.95 0.98 0.96
VGG-19 0 0.97 0.91 0.94 0.95 96.75
1 0.97 0.99 0.98
ViT 0 0.95 0.91 0.93 0.948 96.41
1 0.97 0.98 0.98
ConvMixer 0 0.96 0.95 0.96 0.97 98.0
1 0.98 0.99 0.98

*Class 0: normal (control); Class 1: pneumonia.

Figure S1 shows the pneumonia-infected lungs and normal lungs as detected by the ConvMixer model. The pneumonia classification AUC scores obtained by the LR, k-NN, CNN, VGG-19, ViT, and ConvMixer models, respectively, were 0.87, 0.92, 0.93, 0.95, 0.94, and 0.97 (Figure S2).

4 Discussion

The literature on the recently reported studies about COVID-19 case diagnosis using ML and DL approaches was surveyed for a greater and clearer understanding of the state of the affair. A few selected ones are discussed here. Kanakaprabha and Radha [27] used the CNN method to detect COVID-19 with 95% accuracy. They also detected viral and bacterial pneumonia with 91.46 and 80% accuracy, respectively. Sharma and Tiwari [28] classified COVID-19 and pneumonia using the CXR image database with 94% accuracy. Yaseliani et al. [29] used an ensemble classifier LR and support vector machine with radial basis function to detect pneumonia at 98.55% accuracy. Arias-Garzón et al. [30] achieved 97% detection accuracy for COVID-19 using VGG19 and UNET approaches. Jain et al. [31] used CNN and transfer learning approaches to detect pneumonia from X-ray images and obtained 93.31% accuracy in the CNN model. Moreover, Zhang et al. [32] developed a VGG-based architecture to detect pneumonia and reported accuracy of 96.06, 0.99, 94.408, 90.823, and 92.851%, respectively, for AUC, precision, recall, and F1-score. Mabrouk et al. [33] obtained 93.91% accuracy and 93.88% F1-score employing ensemble learning approaches (Densenet161, mobilenetv2, and ViT). Hashmi et al. [34] proposed the Resnet 50 model to detect pneumonia with 98.14% test accuracy. Recently, Ibrahim et al. [35] used a pre-trained AlexNet model to classify pneumonia and COVID-19 from CXR images with 93.42% accuracy. Makarovskikh and his group [36] classified SARS-CoV-2 positive and normal images using the Densenet121 model with 98.97% accuracy. Our group also developed an IoT-based COVID-19 detection protocol using ML models and obtained 98% accuracy with the k-NN model [21]. Recently, we have also classified COVID-19 and pneumonia from the CXR images by using DL models and obtained 97% accuracy for VGG16 [24].

We have also compared the performance of various models and the state-of-the-art outputs in other similar reported studies, as detailed in Table 6. Using the ensemble model (ResNet18, AlexNet, Inception v3, GoogleNet, and DenseNet121), Chouhan et al. [37] obtained 96.4% accuracy in detecting pneumonia and COVID-19-infected lungs. Using a pre-trained deep CNN model, Liang and Zheng [38] reported 96.7% accuracy in detecting pneumonia in children. The results of the test dataset showed that the recall rate of the method was 96.7% in classifying children with pneumonia, and the F1 score was 92.7%. Using the VGG-16 model, Brunese et al. [39] obtained 96% accuracy, 98% specificity, and 96% sensitivity using the CXR image technique to detect COVID-19 and other pulmonary diseases. Using the CNN model to detect pneumonia from CXR images, Stephen et al. [40] obtained 95% accuracy.

Table 6

Comparison of the observed performances in the study with other reported studies

Literature Dataset % accuracy Class % specificity % sensitivity Reference no.
Chouhan et al. (2020) Public 96.39 3 — — [37]
Liang and Zheng (2020) Public 90 2 — — [38]
Brunese et al. (2020) Public 96 2 98 96 [39]
Stephen et al. (2019) Public 95 2 — — [40]
Wang et al. (2021) Public 89.5 3 0.88 0.87 [41]
Jaiswal et al. (2020) Public 96.25 2 96.21 96.29 [42]
Apostolopoulos and Mpesiana (2020) Public 96.78 3 98.66 96.44 [43]
Proposed model Public 97.09 3 98.4 97.4 This study

Using Inception v3 model to detect COVID-19 and pneumonia-infected lungs, Wang et al. [41] demonstrated 89.5% accuracy, 0.88% specificity, and 0.87% sensitivity. Recently, Jaiswal et al. [42] reported 96.25% accuracy, 96.21% precision, and 96.29% recall using the DenseNet 201 model. Furthermore, Apostolopoulos and Mpesiana [43] obtained 96.78% accuracy, 98.66% specificity, and 96.46 sensitivity using transfer learning approaches to analyse pneumonia, COVID-19, and normal lung CXRs. ConvMixer, the proposed model under the present study, returned the best accuracies (97% for multiclass and 98% for binary class, with combined augmentation techniques) as compared to other similar stand-alone state-of-the-art models.

From the results of the present study, it was observed that the ConvMixer model returned the best accuracies (97% for multiclass and 98% for binary class) through combined augmentation in comparison to other test models. As the proposed model showed greater efficiency in terms of accuracy, specificity, and sensitivity it was recommended to observe to investigate the real-life performance of ConvMixer. Robust evaluation of the efficacy of ConvMixer with its practical utility will aid and contribute significantly to the healthcare sector not only for the benefit of the less-endowed low-income countries but also others. We believe that this will solve the critical challenges in medical diagnosis. It is reported that the SARS-CoV-2 virus was capable of infecting domestic and wild animals, too, through spillback infections [44]. Thus, the present investigation on the predictive models shall help diagnosis and well-being in both humans and animals.

5 Limitations of the study

Although numerous important datasets are freely available online, however, such datasets on CXR images of COVID-19 patients are limited. To overcome this difficulty, only two reliable datasets of CXR images were chosen for investigations in the present study. Both datasets are smaller and have limited COVID-19-related data. Further, the used datasets are reliable but imbalanced. The used datasets had a total of four categories (normal healthy, bacterial, viral, and COVID-19 infected) data. The performance of the proposed model (ConvMixer) must be evaluated to implement in real-life situations by using larger and more categorised datasets.

Some images are of different standards, quality, and sizes, and hence it is highly essential to store the images following the standard operating procedures to allow researchers to utilise the data freely for better image classification. The CXRs are less sensitive than CT scans and may generate false predictions in early and mild cases [45,46]. Therefore, a CT scan could seemingly be more reliable for COVID-19 diagnosis. However, such datasets are limited due to high cost, high radiation dose, and limited resource availability. It is also recommended that the efficacy of the proposed ConvMixer model could be evaluated on CT scan image datasets for further validation.

6 Conclusion

The present work aimed to differentiate and classify healthy lungs (the control) from the COVID-19 and pneumonia-infected lungs based on the CXR images using computer-aided ML and DL models. The viral or bacterial pneumonia-infected lung subtypes were also detected. The ConvMixer model returned detection accuracies of 97.01 and 98% in detecting COVID-19-infected lungs from normal lungs. This model detected bacterial and viral pneumonia variants with 81.72 and 98% accuracies, respectively. After evaluating and comparing five ML/DL models to identify, classify, and categorise COVID-19 and other pneumonia infection forms from the CXR images, ConvMixer is suggested as the best model for COVID-19 detection. The suggested model could accomplish tasks involving three classes and binary classification with overall 97.01, 81.72, and 98% accuracy rates, respectively, for normal, COVID-19, and other pneumonia cases.

The suggested model could be an effective tool to implement in resource-less nations for speedy and reliable diagnoses of the cases affected by COVID-19 or similar infections in the pulmonary and respiratory systems. It will also address the issue of the lack of detection or diagnostic resources in such regions. The suggested ConvMixer network could effectively detect COVID-19 and various other pneumonia using CXR images, which would help radiologists make a timely and accurate diagnosis. The accuracies of these models could further be augmented for diagnosis by using the ensemble techniques.

Acknowledgements

The authors are grateful to their respective Institutions/Universities for the cooperation and support rendered. The authors extend their appreciation to Taif University, Saudi Arabia, for supporting this work through project number (TU-DSPP-2024-20).

  1. Funding information: This research was funded by Taif University, Saudi Arabia, Project No. (TU-DSPP-2024-20).

  2. Author contributions: MP: formal analysis, methodology, and writing – original draft; AKS, ARS, AP, SKB, AYN, IHA, MA, and SMB: data curation and validation and original draft writing; RKM and SM: conceptualisation, project administration, writing – original draft, reviewing and editing.

  3. Conflict of interest: No conflicts of interest to declare.

  4. Data available statement: Datasets used in the study were collected from the open-source repository sites available online as detailed in the “Materials” section. The analysis of the datasets did not require any additional permission.

References

[1] Mohapatra RK, Kuppili S, Suvvari TK, Kandi V, Behera A, Verma S, et al. SARS‐CoV‐2 and its variants of concern including Omicron: a never ending pandemic. Chem Biol Drug Des. 2022;99(5):769–88.10.1111/cbdd.14035Search in Google Scholar PubMed PubMed Central

[2] Dousari AS, Moghadam MT, Satarzadeh N. COVID-19 (Coronavirus disease 2019): A new coronavirus disease. Infect Drug Resist. 2020;13:2819–28.10.2147/IDR.S259279Search in Google Scholar PubMed PubMed Central

[3] Mohapatra RK, Das PK, Kandi V. Challenges in controlling COVID-19 in migrants in Odisha, India. Diabetes Metab Syndr. 2020;14(6):1593–9.10.1016/j.dsx.2020.08.024Search in Google Scholar PubMed PubMed Central

[4] Mohapatra RK, Verma S, Kandi V, Sarangi AK, Seidel V, Das SN, et al. The SARS‐CoV‐2 Omicron variant and its multiple sub‐lineages: transmissibility, vaccine development, antiviral drugs, monoclonal antibodies, and strategies for infection control–a review. ChemistrySelect. 2023;8(9):e202201380.10.1002/slct.202201380Search in Google Scholar

[5] Kutikuppala LS, Kandi V, Sarangi AK, Mishra S, Mohapatra RK. COVID-19 era and the constantly reemerging novel SARS-CoV-2 variants calls for special attention for the geriatrics: a real challenge. Geriatrics. 2022;7(6):143.10.3390/geriatrics7060143Search in Google Scholar PubMed PubMed Central

[6] Mohapatra RK, Kandi V, Gaidhane AM, Zahiruddin QS, Rustagi S, Satapathy P, et al. Global domination of the recently VoI-classified’JN.1′ outcompeting other variants – Comparing the vaccines’ efficacy. ClInfect Pract. 2024;22:100358.10.1016/j.clinpr.2024.100358Search in Google Scholar

[7] Swain RB, Lin X, Wallentin FY. COVID-19 pandemic waves: Identification and interpretation of global data. Heliyon. 2024;10(3):e25090.10.1016/j.heliyon.2024.e25090Search in Google Scholar PubMed PubMed Central

[8] Mohapatra RK, Mishra S, Azam M, Dhama K. COVID-19, WHO guidelines, pedagogy, and respite. Open Med. 2021;16(1):491–3.10.1515/med-2021-0266Search in Google Scholar PubMed PubMed Central

[9] McKibbin W, Fernando R. The global economic impacts of the COVID-19 pandemic. Econ Model. 2023;129:106551.10.1016/j.econmod.2023.106551Search in Google Scholar

[10] Sun M, Yan S, Cao T, Zhang J. The impact of COVID-19 pandemic on the world’s major economies: based on a multi-country and multi-sector CGE model. Front Public Health. 2024;12:1338677.10.3389/fpubh.2024.1338677Search in Google Scholar PubMed PubMed Central

[11] WHO, Impact of COVID-19 on people’s livelihoods, their health and our food systems; October 2020. https://www.who.int/news/item/13-10-2020-impact-of-covid-19-on-people’s-livelihoods-their-health-and-our-food-systems#:∼:text=As%20breadwinners%20lose%20jobs%2C%20fall,indigenous%20peoples%2C%20being%20hardest%20hit. (accessed on 01-11-24).Search in Google Scholar

[12] Peeling RW, Heymann DL, Teo Y-Y, Garcia PJ. Diagnostics for COVID-19: moving from pandemic response to control. Lancet. 2022;399(10326):757–68.10.1016/S0140-6736(21)02346-1Search in Google Scholar PubMed PubMed Central

[13] Xiao AT, Tong YX, Zhang S. Profile of RT-PCR for SARS-CoV-2: A preliminary study from 56 COVID-19 patients. Clin Infect Dis. 2020;71:2249–51.10.1093/cid/ciaa460Search in Google Scholar PubMed PubMed Central

[14] El-Daly MM. Advances and challenges in SARS-CoV-2 detection: A review of molecular and serological technologies. Diagnostics (Basel). 2024;14(5):519.10.3390/diagnostics14050519Search in Google Scholar PubMed PubMed Central

[15] Filchakova O, Dossym D, Ilyas A, Kuanysheva T, Abdizhamil A, Bukasov R. Review of COVID-19 testing and diagnostic methods. Talanta. 2022;244:123409.10.1016/j.talanta.2022.123409Search in Google Scholar PubMed PubMed Central

[16] Thejeshwar SS, Chokkareddy C, Eswaran K. Precise prediction of COVID-19 in chest X-Ray images using KE sieve algorithm. medRxiv [Prepr]. 2020. 10.1101/2020.08.13.20174144.Search in Google Scholar

[17] Absar N, Mamur B, Mahmud A, Emran TB, Khandaker MU, Faruque MRI, et al. Development of a computer-aided tool for detection of COVID-19 pneumonia from CXR images using machine learning algorithm. J Radiat Res Appl Sci. 2022;15(1):32–43.10.1016/j.jrras.2022.02.002Search in Google Scholar

[18] Li L, Qin L, Xu Z, Yin Y, Wang X, Kong B, et al. Using artificial intelligence to detect COVID-19 and community-acquired pneumonia based on pulmonary CT: evaluation of the diagnostic accuracy. Radiology. 2020;296(2):E65–71.10.1148/radiol.2020200905Search in Google Scholar PubMed PubMed Central

[19] Ozsahin I, Sekeroglu B, Musa MS, Mustapha MT, Ozsahin DU. Review on diagnosis of COVID-19 from chest CT images using artificial intelligence. Comput Math Methods Med. 2020;2020:9756518. 10.1155/2020/9756518.Search in Google Scholar PubMed PubMed Central

[20] Baltruschat IM, Nickisch H, Grass M, Knopp T, Saalbach A. Comparison of deep learning approaches for multi-label chest x-ray classification. Sci Rep. 2019;9:6381.10.1038/s41598-019-42294-8Search in Google Scholar PubMed PubMed Central

[21] Pal M, Parija S, Mohapatra RK, Mishra S, Rabaan AA, Al Mutair A, et al. Symptom-based COVID-19 prognosis through AI-based IoT: A bioinformatics approach. Biomed Res Int. 2022;2022:3113119. 10.1155/2022/3113119.Search in Google Scholar PubMed PubMed Central

[22] Peng C-YJ, Lee KL, Ingersoll GM. An introduction to logistic regression analysis and reporting. J Educ Res. 2002;96(1):3–14. 10.1080/00220670209598786.Search in Google Scholar

[23] Pal M, Parija S, Panda G, Dhama K, Mohapatra RK. Risk prediction of cardiovascular disease using machine learning classifiers. Open Med (Wars). 2022;17(1):1100–13.10.1515/med-2022-0508Search in Google Scholar PubMed PubMed Central

[24] Pal M, Parija S, Mishra S, Mohapatra RK, Dhama K. COVID-19 prognosis from chest X-ray images by using deep learning approaches: A next generation diagnostic tool. J Pure Appl Microbiol. 2023;17(2):919–30. 10.22207/JPAM.17.2.20.Search in Google Scholar

[25] Pal M, Panda G, Mohapatra RK, Rath A, Dash S, Shah MA, et al. Ensemble approach of deep learning models for binary and multiclass classification of histopathological images for breast cancer. Pathol Res Pract. 2024;263:155644.10.1016/j.prp.2024.155644Search in Google Scholar PubMed

[26] Cao Z, Huang J, He X, Zong Z. BND-VGG-19: A deep learning algorithm for COVID-19 identification utilizing X-ray images. Knowl Based Syst. 2022;258:110040.10.1016/j.knosys.2022.110040Search in Google Scholar PubMed PubMed Central

[27] Kanakaprabha S, Radha D. Analysis of COVID-19 and pneumonia detection in chest X-ray images using deep learning. 2021 International Conference on Communication, Control and Information Sciences (ICCISc); 2021. 10.1109/ICCISC52257.2021.9484888.Search in Google Scholar

[28] Sharma S, Tiwari S. COVID-19 diagnosis using X-ray images and deep learning. 2021 International Conference on Artificial Intelligence and Smart Systems (ICAIS); 2021. 10.1109/ICAIS50930.2021.9395851.Search in Google Scholar

[29] Yaseliani M, Hamadani AZ, Maghsoodi AI, Mosav A. Pneumonia detection proposing a hybrid deep convolutional neural network based on two parallel visual geometry group architectures and machine learning classifiers. IEEE Access. 2022;10:62110–28.10.1109/ACCESS.2022.3182498Search in Google Scholar

[30] Arias-Garzón D, Alzate-Grisales JA, Orozco-Arias S, Arteaga-Arteaga HB, Bravo-Ortiz MA, Mora-Rubio A, et al. COVID-19 detection in X-ray images using convolutional neural networks. Mach Learn Appl. 2021;6:100138.10.1016/j.mlwa.2021.100138Search in Google Scholar PubMed PubMed Central

[31] Jain R, Nagrath P, Kataria G, Kaushik VS, Hemanth DJ. Pneumonia detection in chest X-ray images using convolutional neural networks and transfer learning. Measurement. 2020;165:108046.10.1016/j.measurement.2020.108046Search in Google Scholar

[32] Zhang D, Ren F, Li Y, Na L, Ma Y. Pneumonia detection from chest X-ray images based on convolutional neural network. Electronics. 2021;10(13):1512.10.3390/electronics10131512Search in Google Scholar

[33] Mabrouk A, Redondo RPD, Dahou A, Elaziz MA, Kayed M. Pneumonia detection on chest X-ray images using ensemble of deep convolutional neural networks. Appl Sci. 2022;12(13):6448.10.3390/app12136448Search in Google Scholar

[34] Hashmi MF, Katiyar S, Hashmic AW, Keska AG. Pneumonia detection in chest X-ray images using compound scaled deep learning model. J Control Meas Electron Comput Commun. 2021;62:397–406.10.1080/00051144.2021.1973297Search in Google Scholar

[35] Ibrahim AU, Ozsoz M, Serte S, AI-turjman F, Yakoi PS. Pneumonia classification using deep learning from chest X-ray images during COVID-19. Cogn Comput. 2024;16:1589–601.10.1007/s12559-020-09787-5Search in Google Scholar PubMed PubMed Central

[36] Makarovskikh T, Salah A, Badr A, Kadi A, Alkattan H, Abotaleb M. Automatic classification Infectious disease X-ray images based on deep learning algorithms. 2022 VIII International Conference on Information Technology and Nanotechnology (ITNT); 2022. 10.1109/ITNT55410.2022.9848538.Search in Google Scholar

[37] Chouhan V, Singh SK, Khamparia A, Gupta D, Tiwari P, Moreira C, et al. A novel transfer learning based approach for pneumonia detection in chest X-ray images. J Appl Sci. 2020;10(2):559.10.3390/app10020559Search in Google Scholar

[38] Liang G, Zheng L. A transfer learning method with deep residual network for pediatric pneumonia diagnosis. Comput Methods Prog Biomed. 2020;187:104964.10.1016/j.cmpb.2019.06.023Search in Google Scholar PubMed

[39] Brunese L, Mercaldo F, Reginelli A, Santone A. Explainable deep learning for pulmonary disease and coronavirus COVID-19 detection from X-rays. Comput Methods Prog Biomed. 2020;196:105608.10.1016/j.cmpb.2020.105608Search in Google Scholar PubMed PubMed Central

[40] Stephen O, Sain M, Maduh UJ, Jeong D-U. An efficient deep learning approach to pneumonia classification in healthcare. J Healthc Eng. 2019;2019:4180949.10.1155/2019/4180949Search in Google Scholar PubMed PubMed Central

[41] Wang S, Kang B, Ma J, Zeng X, Xiao M, Guo J, et al. A deep learning algorithm using CT images to screen for Coronavirus disease (COVID-19). Eur Radiol. 2021;31(8):6096–104.10.1007/s00330-021-07715-1Search in Google Scholar PubMed PubMed Central

[42] Jaiswal A, Gianchandani N, Singh D, Kumar V, Kaur M. Classification of the COVID-19 infected patients using DenseNet201 based deep transfer learning. J Biomol Struct Dyn. 2021;39(15):5682–9.10.1080/07391102.2020.1788642Search in Google Scholar PubMed

[43] Apostolopoulos ID, Mpesiana TA. Covid-19: automatic detection from x-ray images utilizing transfer learning with convolutional neural networks. JPhys Eng Sci Med. 2020;43:635–40.10.1007/s13246-020-00865-4Search in Google Scholar PubMed PubMed Central

[44] Mohapatra RK, Mishra S, Kandi V, Sarangi AK, Ansari A, Chakraborty C, et al. Monkeypox plays a similar role like SARS-CoV-2; intensive animal screening is crucial after the first human-to-dog transmission report–Correspondence. Int J Surg. 2022;106:106925.10.1016/j.ijsu.2022.106925Search in Google Scholar PubMed PubMed Central

[45] El-Din Hemdan E, Shouman MA, Karar ME. Covidx-net: A framework of deep learning classifiers to diagnose covid-19 in x-ray images. arXiv [preprint] arXiv:2003.11055. 2020.Search in Google Scholar

[46] Shyni HM, Chitra E. A comparative study of X-ray and CT images in COVID-19 detection using image processing and deep learning techniques. Comput Methods Prog Biomed Update. 2022;2:100054.10.1016/j.cmpbup.2022.100054Search in Google Scholar PubMed PubMed Central

Received: 2024-07-24
Revised: 2024-11-11
Accepted: 2024-11-17
Published Online: 2025-02-04

© 2025 the author(s), published by De Gruyter

This work is licensed under the Creative Commons Attribution 4.0 International License.

Articles in the same Issue

  1. Research Articles
  2. Network pharmacological analysis and in vitro testing of the rutin effects on triple-negative breast cancer
  3. Impact of diabetes on long-term survival in elderly liver cancer patients: A retrospective study
  4. Knockdown of CCNB1 alleviates high glucose-triggered trophoblast dysfunction during gestational diabetes via Wnt/β-catenin signaling pathway
  5. Risk factors for severe adverse drug reactions in hospitalized patients
  6. Analysis of the effect of ALA-PDT on macrophages in footpad model of mice infected with Fonsecaea monophora based on single-cell sequencing
  7. Development and validation of headspace gas chromatography with a flame ionization detector method for the determination of ethanol in the vitreous humor
  8. CMSP exerts anti-tumor effects on small cell lung cancer cells by inducing mitochondrial dysfunction and ferroptosis
  9. Predictive value of plasma sB7-H3 and YKL-40 in pediatric refractory Mycoplasma pneumoniae pneumonia
  10. Antiangiogenic potential of Elaeagnus umbellata extracts and molecular docking study by targeting VEGFR-2 pathway
  11. Comparison of the effectiveness of nurse-led preoperative counseling and postoperative follow-up care vs standard care for patients with gastric cancer
  12. Comparing the therapeutic efficacy of endoscopic minimally invasive surgery and traditional surgery for early-stage breast cancer: A meta-analysis
  13. Adhered macrophages as an additional marker of cardiomyocyte injury in biopsies of patients with dilated cardiomyopathy
  14. Association between statin administration and outcome in patients with sepsis: A retrospective study
  15. Exploration of the association between estimated glucose disposal rate and osteoarthritis in middle-aged and older adults: An analysis of NHANES data from 2011 to 2018
  16. A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models
  17. Lysophosphatidic acid 2 alleviates deep vein thrombosis via protective endothelial barrier function
  18. Transcription factor A, mitochondrial promotes lymph node metastasis and lymphangiogenesis in epithelial ovarian carcinoma
  19. Serum PM20D1 levels are associated with nutritional status and inflammatory factors in gastric cancer patients undergoing early enteral nutrition
  20. Hydromorphone reduced the incidence of emergence agitation after adenotonsillectomy in children with obstructive sleep apnea: A randomized, double-blind study
  21. Vitamin D replacement therapy may regulate sleep habits in patients with restless leg syndrome
  22. The first-line antihypertensive nitrendipine potentiated the therapeutic effect of oxaliplatin by downregulating CACNA1D in colorectal cancer
  23. Health literacy and health-related quality of life: The mediating role of irrational happiness
  24. Modulatory effects of Lycium barbarum polysaccharide on bone cell dynamics in osteoporosis
  25. Mechanism research on inhibition of gastric cancer in vitro by the extract of Pinellia ternata based on network pharmacology and cellular metabolomics
  26. Examination of the causal role of immune cells in non-alcoholic fatty liver disease by a bidirectional Mendelian randomization study
  27. Clinical analysis of ten cases of HIV infection combined with acute leukemia
  28. Investigating the cardioprotective potential of quercetin against tacrolimus-induced cardiotoxicity in Wistar rats: A mechanistic insights
  29. Clinical observation of probiotics combined with mesalazine and Yiyi Baitouweng Decoction retention enema in treating mild-to-moderate ulcerative colitis
  30. Diagnostic value of ratio of blood inflammation to coagulation markers in periprosthetic joint infection
  31. Sex-specific associations of sex hormone binding globulin and risk of bladder cancer
  32. Core muscle strength and stability-oriented breathing training reduces inter-recti distance in postpartum women
  33. The ERAS nursing care strategy for patients undergoing transsphenoidal endoscopic pituitary tumor resection: A randomized blinded controlled trial
  34. The serum IL-17A levels in patients with traumatic bowel rupture post-surgery and its predictive value for patient prognosis
  35. Impact of Kolb’s experiential learning theory-based nursing on caregiver burden and psychological state of caregivers of dementia patients
  36. Analysis of serum NLR combined with intraoperative margin condition to predict the prognosis of cervical HSIL patients undergoing LEEP surgery
  37. Commiphora gileadensis ameliorate infertility and erectile dysfunction in diabetic male mice
  38. The correlation between epithelial–mesenchymal transition classification and MMP2 expression of circulating tumor cells and prognosis of advanced or metastatic nasopharyngeal carcinoma
  39. Tetrahydropalmatine improves mitochondrial function in vascular smooth muscle cells of atherosclerosis in vitro by inhibiting Ras homolog gene family A/Rho-associated protein kinase-1 signaling pathway
  40. A cross-sectional study: Relationship between serum oxidative stress levels and arteriovenous fistula maturation in maintenance dialysis patients
  41. A comparative analysis of the impact of repeated administration of flavan 3-ol on brown, subcutaneous, and visceral adipose tissue
  42. Identifying early screening factors for depression in middle-aged and older adults: A cohort study
  43. Perform tumor-specific survival analysis for Merkel cell carcinoma patients undergoing surgical resection based on the SEER database by constructing a nomogram chart
  44. Unveiling the role of CXCL10 in pancreatic cancer progression: A novel prognostic indicator
  45. High-dose preoperative intraperitoneal erythropoietin and intravenous methylprednisolone in acute traumatic spinal cord injuries following decompression surgeries
  46. RAB39B: A novel biomarker for acute myeloid leukemia identified via multi-omics and functional validation
  47. Impact of peripheral conditioning on reperfusion injury following primary percutaneous coronary intervention in diabetic and non-diabetic STEMI patients
  48. Clinical efficacy of azacitidine in the treatment of middle- and high-risk myelodysplastic syndrome in middle-aged and elderly patients: A retrospective study
  49. The effect of ambulatory blood pressure load on mitral regurgitation in continuous ambulatory peritoneal dialysis patients
  50. Expression and clinical significance of ITGA3 in breast cancer
  51. Single-nucleus RNA sequencing reveals ARHGAP28 expression of podocytes as a biomarker in human diabetic nephropathy
  52. rSIG combined with NLR in the prognostic assessment of patients with multiple injuries
  53. Toxic metals and metalloids in collagen supplements of fish and jellyfish origin: Risk assessment for daily intake
  54. Exploring causal relationship between 41 inflammatory cytokines and marginal zone lymphoma: A bidirectional Mendelian randomization study
  55. Gender beliefs and legitimization of dating violence in adolescents
  56. Effect of serum IL-6, CRP, and MMP-9 levels on the efficacy of modified preperitoneal Kugel repair in patients with inguinal hernia
  57. Effect of smoking and smoking cessation on hematological parameters in polycythemic patients
  58. Pathogen surveillance and risk factors for pulmonary infection in patients with lung cancer: A retrospective single-center study
  59. Necroptosis of hippocampal neurons in paclitaxel chemotherapy-induced cognitive impairment mediates microglial activation via TLR4/MyD88 signaling pathway
  60. Celastrol suppresses neovascularization in rat aortic vascular endothelial cells stimulated by inflammatory tenocytes via modulating the NLRP3 pathway
  61. Cord-lamina angle and foraminal diameter as key predictors of C5 palsy after anterior cervical decompression and fusion surgery
  62. GATA1: A key biomarker for predicting the prognosis of patients with diffuse large B-cell lymphoma
  63. Influencing factors of false lumen thrombosis in type B aortic dissection: A single-center retrospective study
  64. MZB1 regulates the immune microenvironment and inhibits ovarian cancer cell migration
  65. Integrating experimental and network pharmacology to explore the pharmacological mechanisms of Dioscin against glioblastoma
  66. Trends in research on preterm birth in twin pregnancy based on bibliometrics
  67. Four-week IgE/baseline IgE ratio combined with tryptase predicts clinical outcome in omalizumab-treated children with moderate-to-severe asthma
  68. Single-cell transcriptomic analysis identifies a stress response Schwann cell subtype
  69. Acute pancreatitis risk in the diagnosis and management of inflammatory bowel disease: A critical focus
  70. Effect of subclinical esketamine on NLRP3 and cognitive dysfunction in elderly ischemic stroke patients
  71. Interleukin-37 mediates the anti-oral tumor activity in oral cancer through STAT3
  72. CA199 and CEA expression levels, and minimally invasive postoperative prognosis analysis in esophageal squamous carcinoma patients
  73. Efficacy of a novel drainage catheter in the treatment of CSF leak after posterior spine surgery: A retrospective cohort study
  74. Comprehensive biomedicine assessment of Apteranthes tuberculata extracts: Phytochemical analysis and multifaceted pharmacological evaluation in animal models
  75. Relation of time in range to severity of coronary artery disease in patients with type 2 diabetes: A cross-sectional study
  76. Dopamine attenuates ethanol-induced neuronal apoptosis by stimulating electrical activity in the developing rat retina
  77. Correlation between albumin levels during the third trimester and the risk of postpartum levator ani muscle rupture
  78. Factors associated with maternal attention and distraction during breastfeeding and childcare: A cross-sectional study in the west of Iran
  79. Mechanisms of hesperetin in treating metabolic dysfunction-associated steatosis liver disease via network pharmacology and in vitro experiments
  80. The law on oncological oblivion in the Italian and European context: How to best uphold the cancer patients’ rights to privacy and self-determination?
  81. The prognostic value of the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, and prognostic nutritional index for survival in patients with colorectal cancer
  82. Factors affecting the measurements of peripheral oxygen saturation values in healthy young adults
  83. Comparison and correlations between findings of hysteroscopy and vaginal color Doppler ultrasonography for detection of uterine abnormalities in patients with recurrent implantation failure
  84. The effects of different types of RAGT on balance function in stroke patients with low levels of independent walking in a convalescent rehabilitation hospital
  85. Causal relationship between asthma and ankylosing spondylitis: A bidirectional two-sample univariable and multivariable Mendelian randomization study
  86. Correlations of health literacy with individuals’ understanding and use of medications in Southern Taiwan
  87. Correlation of serum calprotectin with outcome of acute cerebral infarction
  88. Comparison of computed tomography and guided bronchoscopy in the diagnosis of pulmonary nodules: A systematic review and meta-analysis
  89. Curdione protects vascular endothelial cells and atherosclerosis via the regulation of DNMT1-mediated ERBB4 promoter methylation
  90. The identification of novel missense variant in ChAT gene in a patient with gestational diabetes denotes plausible genetic association
  91. Molecular genotyping of multi-system rare blood types in foreign blood donors based on DNA sequencing and its clinical significance
  92. Exploring the role of succinyl carnitine in the association between CD39⁺ CD4⁺ T cell and ulcerative colitis: A Mendelian randomization study
  93. Dexmedetomidine suppresses microglial activation in postoperative cognitive dysfunction via the mmu-miRNA-125/TRAF6 signaling axis
  94. Analysis of serum metabolomics in patients with different types of chronic heart failure
  95. Diagnostic value of hematological parameters in the early diagnosis of acute cholecystitis
  96. Pachymaran alleviates fat accumulation, hepatocyte degeneration, and injury in mice with nonalcoholic fatty liver disease
  97. Decrease in CD4 and CD8 lymphocytes are predictors of severe clinical picture and unfavorable outcome of the disease in patients with COVID-19
  98. METTL3 blocked the progression of diabetic retinopathy through m6A-modified SOX2
  99. The predictive significance of anti-RO-52 antibody in patients with interstitial pneumonia after treatment of malignant tumors
  100. Exploring cerebrospinal fluid metabolites, cognitive function, and brain atrophy: Insights from Mendelian randomization
  101. Development and validation of potential molecular subtypes and signatures of ocular sarcoidosis based on autophagy-related gene analysis
  102. Widespread venous thrombosis: Unveiling a complex case of Behçet’s disease with a literature perspective
  103. Uterine fibroid embolization: An analysis of clinical outcomes and impact on patients’ quality of life
  104. Discovery of lipid metabolism-related diagnostic biomarkers and construction of diagnostic model in steroid-induced osteonecrosis of femoral head
  105. Serum-derived exomiR-188-3p is a promising novel biomarker for early-stage ovarian cancer
  106. Enhancing chronic back pain management: A comparative study of ultrasound–MRI fusion guidance for paravertebral nerve block
  107. Peptide CCAT1-70aa promotes hepatocellular carcinoma proliferation and invasion via the MAPK/ERK pathway
  108. Electroacupuncture-induced reduction of myocardial ischemia–reperfusion injury via FTO-dependent m6A methylation modulation
  109. Hemorrhoids and cardiovascular disease: A bidirectional Mendelian randomization study
  110. Cell-free adipose extract inhibits hypertrophic scar formation through collagen remodeling and antiangiogenesis
  111. HALP score in Demodex blepharitis: A case–control study
  112. Assessment of SOX2 performance as a marker for circulating cancer stem-like cells (CCSCs) identification in advanced breast cancer patients using CytoTrack system
  113. Risk and prognosis for brain metastasis in primary metastatic cervical cancer patients: A population-based study
  114. Comparison of the two intestinal anastomosis methods in pediatric patients
  115. Factors influencing hematological toxicity and adverse effects of perioperative hyperthermic intraperitoneal vs intraperitoneal chemotherapy in gastrointestinal cancer
  116. Endotoxin tolerance inhibits NLRP3 inflammasome activation in macrophages of septic mice by restoring autophagic flux through TRIM26
  117. Review Articles
  118. The effects of enhanced external counter-pulsation on post-acute sequelae of COVID-19: A narrative review
  119. Diabetes-related cognitive impairment: Mechanisms, symptoms, and treatments
  120. Microscopic changes and gross morphology of placenta in women affected by gestational diabetes mellitus in dietary treatment: A systematic review
  121. Review of mechanisms and frontier applications in IL-17A-induced hypertension
  122. Research progress on the correlation between islet amyloid peptides and type 2 diabetes mellitus
  123. The safety and efficacy of BCG combined with mitomycin C compared with BCG monotherapy in patients with non-muscle-invasive bladder cancer: A systematic review and meta-analysis
  124. The application of augmented reality in robotic general surgery: A mini-review
  125. The effect of Greek mountain tea extract and wheat germ extract on peripheral blood flow and eicosanoid metabolism in mammals
  126. Neurogasobiology of migraine: Carbon monoxide, hydrogen sulfide, and nitric oxide as emerging pathophysiological trinacrium relevant to nociception regulation
  127. Plant polyphenols, terpenes, and terpenoids in oral health
  128. Laboratory medicine between technological innovation, rights safeguarding, and patient safety: A bioethical perspective
  129. End-of-life in cancer patients: Medicolegal implications and ethical challenges in Europe
  130. The maternal factors during pregnancy for intrauterine growth retardation: An umbrella review
  131. Intra-abdominal hypertension/abdominal compartment syndrome of pediatric patients in critical care settings
  132. PI3K/Akt pathway and neuroinflammation in sepsis-associated encephalopathy
  133. Screening of Group B Streptococcus in pregnancy: A systematic review for the laboratory detection
  134. Giant borderline ovarian tumours – review of the literature
  135. Leveraging artificial intelligence for collaborative care planning: Innovations and impacts in shared decision-making – A systematic review
  136. Cholera epidemiology analysis through the experience of the 1973 Naples epidemic
  137. Case Reports
  138. Delayed graft function after renal transplantation
  139. Semaglutide treatment for type 2 diabetes in a patient with chronic myeloid leukemia: A case report and review of the literature
  140. Diverse electrophysiological demyelinating features in a late-onset glycogen storage disease type IIIa case
  141. Giant right atrial hemangioma presenting with ascites: A case report
  142. Laser excision of a large granular cell tumor of the vocal cord with subglottic extension: A case report
  143. Rapid Communication
  144. Biological properties of valve materials using RGD and EC
  145. Letter to the Editor
  146. Role of enhanced external counterpulsation in long COVID
  147. Expression of Concern
  148. Expression of concern “A ceRNA network mediated by LINC00475 in papillary thyroid carcinoma”
  149. Expression of concern “Notoginsenoside R1 alleviates spinal cord injury through the miR-301a/KLF7 axis to activate Wnt/β-catenin pathway”
  150. Expression of concern “circ_0020123 promotes cell proliferation and migration in lung adenocarcinoma via PDZD8”
  151. Corrigendum
  152. Corrigendum to “Empagliflozin improves aortic injury in obese mice by regulating fatty acid metabolism”
  153. Corrigendum to “Comparing the therapeutic efficacy of endoscopic minimally invasive surgery and traditional surgery for early-stage breast cancer: A meta-analysis”
  154. Corrigendum to “The progress of autoimmune hepatitis research and future challenges”
  155. Retraction
  156. Retraction of “miR-654-5p promotes gastric cancer progression via the GPRIN1/NF-κB pathway”
  157. Special Issue Advancements in oncology: bridging clinical and experimental research - Part II
  158. Unveiling novel biomarkers for platinum chemoresistance in ovarian cancer
  159. Lathyrol affects the expression of AR and PSA and inhibits the malignant behavior of RCC cells
  160. The era of increasing cancer survivorship: Trends in fertility preservation, medico-legal implications, and ethical challenges
  161. Bone scintigraphy and positron emission tomography in the early diagnosis of MRONJ
  162. Meta-analysis of clinical efficacy and safety of immunotherapy combined with chemotherapy in non-small cell lung cancer
  163. Special Issue Computational Intelligence Methodologies Meets Recurrent Cancers - Part IV
  164. Exploration of mRNA-modifying METTL3 oncogene as momentous prognostic biomarker responsible for colorectal cancer development
  165. Special Issue The evolving saga of RNAs from bench to bedside - Part III
  166. Interaction and verification of ferroptosis-related RNAs Rela and Stat3 in promoting sepsis-associated acute kidney injury
  167. Special Issue Exploring the biological mechanism of human diseases based on MultiOmics Technology - Part II
  168. Dynamic changes in lactate-related genes in microglia and their role in immune cell interactions after ischemic stroke
  169. A prognostic model correlated with fatty acid metabolism in Ewing’s sarcoma based on bioinformatics analysis
  170. Special Issue Diabetes
  171. Nutritional risk assessment and nutritional support in children with congenital diabetes during surgery
  172. Correlation of the differential expressions of RANK, RANKL, and OPG with obesity in the elderly population in Xinjiang
  173. Special Issue Biomarker Discovery and Precision Medicine
  174. CircASH1L-mediated tumor progression in triple-negative breast cancer: PI3K/AKT pathway mechanisms
https://doi.org/10.1515/med-2024-1110
Keywords for this article
COVID-19; chest X-ray images; ML and DL models; binary and multiclass classification
Creative Commons
BY 4.0

Articles in the same Issue

  1. Research Articles
  2. Network pharmacological analysis and in vitro testing of the rutin effects on triple-negative breast cancer
  3. Impact of diabetes on long-term survival in elderly liver cancer patients: A retrospective study
  4. Knockdown of CCNB1 alleviates high glucose-triggered trophoblast dysfunction during gestational diabetes via Wnt/β-catenin signaling pathway
  5. Risk factors for severe adverse drug reactions in hospitalized patients
  6. Analysis of the effect of ALA-PDT on macrophages in footpad model of mice infected with Fonsecaea monophora based on single-cell sequencing
  7. Development and validation of headspace gas chromatography with a flame ionization detector method for the determination of ethanol in the vitreous humor
  8. CMSP exerts anti-tumor effects on small cell lung cancer cells by inducing mitochondrial dysfunction and ferroptosis
  9. Predictive value of plasma sB7-H3 and YKL-40 in pediatric refractory Mycoplasma pneumoniae pneumonia
  10. Antiangiogenic potential of Elaeagnus umbellata extracts and molecular docking study by targeting VEGFR-2 pathway
  11. Comparison of the effectiveness of nurse-led preoperative counseling and postoperative follow-up care vs standard care for patients with gastric cancer
  12. Comparing the therapeutic efficacy of endoscopic minimally invasive surgery and traditional surgery for early-stage breast cancer: A meta-analysis
  13. Adhered macrophages as an additional marker of cardiomyocyte injury in biopsies of patients with dilated cardiomyopathy
  14. Association between statin administration and outcome in patients with sepsis: A retrospective study
  15. Exploration of the association between estimated glucose disposal rate and osteoarthritis in middle-aged and older adults: An analysis of NHANES data from 2011 to 2018
  16. A comparative analysis of the binary and multiclass classified chest X-ray images of pneumonia and COVID-19 with ML and DL models
  17. Lysophosphatidic acid 2 alleviates deep vein thrombosis via protective endothelial barrier function
  18. Transcription factor A, mitochondrial promotes lymph node metastasis and lymphangiogenesis in epithelial ovarian carcinoma
  19. Serum PM20D1 levels are associated with nutritional status and inflammatory factors in gastric cancer patients undergoing early enteral nutrition
  20. Hydromorphone reduced the incidence of emergence agitation after adenotonsillectomy in children with obstructive sleep apnea: A randomized, double-blind study
  21. Vitamin D replacement therapy may regulate sleep habits in patients with restless leg syndrome
  22. The first-line antihypertensive nitrendipine potentiated the therapeutic effect of oxaliplatin by downregulating CACNA1D in colorectal cancer
  23. Health literacy and health-related quality of life: The mediating role of irrational happiness
  24. Modulatory effects of Lycium barbarum polysaccharide on bone cell dynamics in osteoporosis
  25. Mechanism research on inhibition of gastric cancer in vitro by the extract of Pinellia ternata based on network pharmacology and cellular metabolomics
  26. Examination of the causal role of immune cells in non-alcoholic fatty liver disease by a bidirectional Mendelian randomization study
  27. Clinical analysis of ten cases of HIV infection combined with acute leukemia
  28. Investigating the cardioprotective potential of quercetin against tacrolimus-induced cardiotoxicity in Wistar rats: A mechanistic insights
  29. Clinical observation of probiotics combined with mesalazine and Yiyi Baitouweng Decoction retention enema in treating mild-to-moderate ulcerative colitis
  30. Diagnostic value of ratio of blood inflammation to coagulation markers in periprosthetic joint infection
  31. Sex-specific associations of sex hormone binding globulin and risk of bladder cancer
  32. Core muscle strength and stability-oriented breathing training reduces inter-recti distance in postpartum women
  33. The ERAS nursing care strategy for patients undergoing transsphenoidal endoscopic pituitary tumor resection: A randomized blinded controlled trial
  34. The serum IL-17A levels in patients with traumatic bowel rupture post-surgery and its predictive value for patient prognosis
  35. Impact of Kolb’s experiential learning theory-based nursing on caregiver burden and psychological state of caregivers of dementia patients
  36. Analysis of serum NLR combined with intraoperative margin condition to predict the prognosis of cervical HSIL patients undergoing LEEP surgery
  37. Commiphora gileadensis ameliorate infertility and erectile dysfunction in diabetic male mice
  38. The correlation between epithelial–mesenchymal transition classification and MMP2 expression of circulating tumor cells and prognosis of advanced or metastatic nasopharyngeal carcinoma
  39. Tetrahydropalmatine improves mitochondrial function in vascular smooth muscle cells of atherosclerosis in vitro by inhibiting Ras homolog gene family A/Rho-associated protein kinase-1 signaling pathway
  40. A cross-sectional study: Relationship between serum oxidative stress levels and arteriovenous fistula maturation in maintenance dialysis patients
  41. A comparative analysis of the impact of repeated administration of flavan 3-ol on brown, subcutaneous, and visceral adipose tissue
  42. Identifying early screening factors for depression in middle-aged and older adults: A cohort study
  43. Perform tumor-specific survival analysis for Merkel cell carcinoma patients undergoing surgical resection based on the SEER database by constructing a nomogram chart
  44. Unveiling the role of CXCL10 in pancreatic cancer progression: A novel prognostic indicator
  45. High-dose preoperative intraperitoneal erythropoietin and intravenous methylprednisolone in acute traumatic spinal cord injuries following decompression surgeries
  46. RAB39B: A novel biomarker for acute myeloid leukemia identified via multi-omics and functional validation
  47. Impact of peripheral conditioning on reperfusion injury following primary percutaneous coronary intervention in diabetic and non-diabetic STEMI patients
  48. Clinical efficacy of azacitidine in the treatment of middle- and high-risk myelodysplastic syndrome in middle-aged and elderly patients: A retrospective study
  49. The effect of ambulatory blood pressure load on mitral regurgitation in continuous ambulatory peritoneal dialysis patients
  50. Expression and clinical significance of ITGA3 in breast cancer
  51. Single-nucleus RNA sequencing reveals ARHGAP28 expression of podocytes as a biomarker in human diabetic nephropathy
  52. rSIG combined with NLR in the prognostic assessment of patients with multiple injuries
  53. Toxic metals and metalloids in collagen supplements of fish and jellyfish origin: Risk assessment for daily intake
  54. Exploring causal relationship between 41 inflammatory cytokines and marginal zone lymphoma: A bidirectional Mendelian randomization study
  55. Gender beliefs and legitimization of dating violence in adolescents
  56. Effect of serum IL-6, CRP, and MMP-9 levels on the efficacy of modified preperitoneal Kugel repair in patients with inguinal hernia
  57. Effect of smoking and smoking cessation on hematological parameters in polycythemic patients
  58. Pathogen surveillance and risk factors for pulmonary infection in patients with lung cancer: A retrospective single-center study
  59. Necroptosis of hippocampal neurons in paclitaxel chemotherapy-induced cognitive impairment mediates microglial activation via TLR4/MyD88 signaling pathway
  60. Celastrol suppresses neovascularization in rat aortic vascular endothelial cells stimulated by inflammatory tenocytes via modulating the NLRP3 pathway
  61. Cord-lamina angle and foraminal diameter as key predictors of C5 palsy after anterior cervical decompression and fusion surgery
  62. GATA1: A key biomarker for predicting the prognosis of patients with diffuse large B-cell lymphoma
  63. Influencing factors of false lumen thrombosis in type B aortic dissection: A single-center retrospective study
  64. MZB1 regulates the immune microenvironment and inhibits ovarian cancer cell migration
  65. Integrating experimental and network pharmacology to explore the pharmacological mechanisms of Dioscin against glioblastoma
  66. Trends in research on preterm birth in twin pregnancy based on bibliometrics
  67. Four-week IgE/baseline IgE ratio combined with tryptase predicts clinical outcome in omalizumab-treated children with moderate-to-severe asthma
  68. Single-cell transcriptomic analysis identifies a stress response Schwann cell subtype
  69. Acute pancreatitis risk in the diagnosis and management of inflammatory bowel disease: A critical focus
  70. Effect of subclinical esketamine on NLRP3 and cognitive dysfunction in elderly ischemic stroke patients
  71. Interleukin-37 mediates the anti-oral tumor activity in oral cancer through STAT3
  72. CA199 and CEA expression levels, and minimally invasive postoperative prognosis analysis in esophageal squamous carcinoma patients
  73. Efficacy of a novel drainage catheter in the treatment of CSF leak after posterior spine surgery: A retrospective cohort study
  74. Comprehensive biomedicine assessment of Apteranthes tuberculata extracts: Phytochemical analysis and multifaceted pharmacological evaluation in animal models
  75. Relation of time in range to severity of coronary artery disease in patients with type 2 diabetes: A cross-sectional study
  76. Dopamine attenuates ethanol-induced neuronal apoptosis by stimulating electrical activity in the developing rat retina
  77. Correlation between albumin levels during the third trimester and the risk of postpartum levator ani muscle rupture
  78. Factors associated with maternal attention and distraction during breastfeeding and childcare: A cross-sectional study in the west of Iran
  79. Mechanisms of hesperetin in treating metabolic dysfunction-associated steatosis liver disease via network pharmacology and in vitro experiments
  80. The law on oncological oblivion in the Italian and European context: How to best uphold the cancer patients’ rights to privacy and self-determination?
  81. The prognostic value of the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, and prognostic nutritional index for survival in patients with colorectal cancer
  82. Factors affecting the measurements of peripheral oxygen saturation values in healthy young adults
  83. Comparison and correlations between findings of hysteroscopy and vaginal color Doppler ultrasonography for detection of uterine abnormalities in patients with recurrent implantation failure
  84. The effects of different types of RAGT on balance function in stroke patients with low levels of independent walking in a convalescent rehabilitation hospital
  85. Causal relationship between asthma and ankylosing spondylitis: A bidirectional two-sample univariable and multivariable Mendelian randomization study
  86. Correlations of health literacy with individuals’ understanding and use of medications in Southern Taiwan
  87. Correlation of serum calprotectin with outcome of acute cerebral infarction
  88. Comparison of computed tomography and guided bronchoscopy in the diagnosis of pulmonary nodules: A systematic review and meta-analysis
  89. Curdione protects vascular endothelial cells and atherosclerosis via the regulation of DNMT1-mediated ERBB4 promoter methylation
  90. The identification of novel missense variant in ChAT gene in a patient with gestational diabetes denotes plausible genetic association
  91. Molecular genotyping of multi-system rare blood types in foreign blood donors based on DNA sequencing and its clinical significance
  92. Exploring the role of succinyl carnitine in the association between CD39⁺ CD4⁺ T cell and ulcerative colitis: A Mendelian randomization study
  93. Dexmedetomidine suppresses microglial activation in postoperative cognitive dysfunction via the mmu-miRNA-125/TRAF6 signaling axis
  94. Analysis of serum metabolomics in patients with different types of chronic heart failure
  95. Diagnostic value of hematological parameters in the early diagnosis of acute cholecystitis
  96. Pachymaran alleviates fat accumulation, hepatocyte degeneration, and injury in mice with nonalcoholic fatty liver disease
  97. Decrease in CD4 and CD8 lymphocytes are predictors of severe clinical picture and unfavorable outcome of the disease in patients with COVID-19
  98. METTL3 blocked the progression of diabetic retinopathy through m6A-modified SOX2
  99. The predictive significance of anti-RO-52 antibody in patients with interstitial pneumonia after treatment of malignant tumors
  100. Exploring cerebrospinal fluid metabolites, cognitive function, and brain atrophy: Insights from Mendelian randomization
  101. Development and validation of potential molecular subtypes and signatures of ocular sarcoidosis based on autophagy-related gene analysis
  102. Widespread venous thrombosis: Unveiling a complex case of Behçet’s disease with a literature perspective
  103. Uterine fibroid embolization: An analysis of clinical outcomes and impact on patients’ quality of life
  104. Discovery of lipid metabolism-related diagnostic biomarkers and construction of diagnostic model in steroid-induced osteonecrosis of femoral head
  105. Serum-derived exomiR-188-3p is a promising novel biomarker for early-stage ovarian cancer
  106. Enhancing chronic back pain management: A comparative study of ultrasound–MRI fusion guidance for paravertebral nerve block
  107. Peptide CCAT1-70aa promotes hepatocellular carcinoma proliferation and invasion via the MAPK/ERK pathway
  108. Electroacupuncture-induced reduction of myocardial ischemia–reperfusion injury via FTO-dependent m6A methylation modulation
  109. Hemorrhoids and cardiovascular disease: A bidirectional Mendelian randomization study
  110. Cell-free adipose extract inhibits hypertrophic scar formation through collagen remodeling and antiangiogenesis
  111. HALP score in Demodex blepharitis: A case–control study
  112. Assessment of SOX2 performance as a marker for circulating cancer stem-like cells (CCSCs) identification in advanced breast cancer patients using CytoTrack system
  113. Risk and prognosis for brain metastasis in primary metastatic cervical cancer patients: A population-based study
  114. Comparison of the two intestinal anastomosis methods in pediatric patients
  115. Factors influencing hematological toxicity and adverse effects of perioperative hyperthermic intraperitoneal vs intraperitoneal chemotherapy in gastrointestinal cancer
  116. Endotoxin tolerance inhibits NLRP3 inflammasome activation in macrophages of septic mice by restoring autophagic flux through TRIM26
  117. Review Articles
  118. The effects of enhanced external counter-pulsation on post-acute sequelae of COVID-19: A narrative review
  119. Diabetes-related cognitive impairment: Mechanisms, symptoms, and treatments
  120. Microscopic changes and gross morphology of placenta in women affected by gestational diabetes mellitus in dietary treatment: A systematic review
  121. Review of mechanisms and frontier applications in IL-17A-induced hypertension
  122. Research progress on the correlation between islet amyloid peptides and type 2 diabetes mellitus
  123. The safety and efficacy of BCG combined with mitomycin C compared with BCG monotherapy in patients with non-muscle-invasive bladder cancer: A systematic review and meta-analysis
  124. The application of augmented reality in robotic general surgery: A mini-review
  125. The effect of Greek mountain tea extract and wheat germ extract on peripheral blood flow and eicosanoid metabolism in mammals
  126. Neurogasobiology of migraine: Carbon monoxide, hydrogen sulfide, and nitric oxide as emerging pathophysiological trinacrium relevant to nociception regulation
  127. Plant polyphenols, terpenes, and terpenoids in oral health
  128. Laboratory medicine between technological innovation, rights safeguarding, and patient safety: A bioethical perspective
  129. End-of-life in cancer patients: Medicolegal implications and ethical challenges in Europe
  130. The maternal factors during pregnancy for intrauterine growth retardation: An umbrella review
  131. Intra-abdominal hypertension/abdominal compartment syndrome of pediatric patients in critical care settings
  132. PI3K/Akt pathway and neuroinflammation in sepsis-associated encephalopathy
  133. Screening of Group B Streptococcus in pregnancy: A systematic review for the laboratory detection
  134. Giant borderline ovarian tumours – review of the literature
  135. Leveraging artificial intelligence for collaborative care planning: Innovations and impacts in shared decision-making – A systematic review
  136. Cholera epidemiology analysis through the experience of the 1973 Naples epidemic
  137. Case Reports
  138. Delayed graft function after renal transplantation
  139. Semaglutide treatment for type 2 diabetes in a patient with chronic myeloid leukemia: A case report and review of the literature
  140. Diverse electrophysiological demyelinating features in a late-onset glycogen storage disease type IIIa case
  141. Giant right atrial hemangioma presenting with ascites: A case report
  142. Laser excision of a large granular cell tumor of the vocal cord with subglottic extension: A case report
  143. Rapid Communication
  144. Biological properties of valve materials using RGD and EC
  145. Letter to the Editor
  146. Role of enhanced external counterpulsation in long COVID
  147. Expression of Concern
  148. Expression of concern “A ceRNA network mediated by LINC00475 in papillary thyroid carcinoma”
  149. Expression of concern “Notoginsenoside R1 alleviates spinal cord injury through the miR-301a/KLF7 axis to activate Wnt/β-catenin pathway”
  150. Expression of concern “circ_0020123 promotes cell proliferation and migration in lung adenocarcinoma via PDZD8”
  151. Corrigendum
  152. Corrigendum to “Empagliflozin improves aortic injury in obese mice by regulating fatty acid metabolism”
  153. Corrigendum to “Comparing the therapeutic efficacy of endoscopic minimally invasive surgery and traditional surgery for early-stage breast cancer: A meta-analysis”
  154. Corrigendum to “The progress of autoimmune hepatitis research and future challenges”
  155. Retraction
  156. Retraction of “miR-654-5p promotes gastric cancer progression via the GPRIN1/NF-κB pathway”
  157. Special Issue Advancements in oncology: bridging clinical and experimental research - Part II
  158. Unveiling novel biomarkers for platinum chemoresistance in ovarian cancer
  159. Lathyrol affects the expression of AR and PSA and inhibits the malignant behavior of RCC cells
  160. The era of increasing cancer survivorship: Trends in fertility preservation, medico-legal implications, and ethical challenges
  161. Bone scintigraphy and positron emission tomography in the early diagnosis of MRONJ
  162. Meta-analysis of clinical efficacy and safety of immunotherapy combined with chemotherapy in non-small cell lung cancer
  163. Special Issue Computational Intelligence Methodologies Meets Recurrent Cancers - Part IV
  164. Exploration of mRNA-modifying METTL3 oncogene as momentous prognostic biomarker responsible for colorectal cancer development
  165. Special Issue The evolving saga of RNAs from bench to bedside - Part III
  166. Interaction and verification of ferroptosis-related RNAs Rela and Stat3 in promoting sepsis-associated acute kidney injury
  167. Special Issue Exploring the biological mechanism of human diseases based on MultiOmics Technology - Part II
  168. Dynamic changes in lactate-related genes in microglia and their role in immune cell interactions after ischemic stroke
  169. A prognostic model correlated with fatty acid metabolism in Ewing’s sarcoma based on bioinformatics analysis
  170. Special Issue Diabetes
  171. Nutritional risk assessment and nutritional support in children with congenital diabetes during surgery
  172. Correlation of the differential expressions of RANK, RANKL, and OPG with obesity in the elderly population in Xinjiang
  173. Special Issue Biomarker Discovery and Precision Medicine
  174. CircASH1L-mediated tumor progression in triple-negative breast cancer: PI3K/AKT pathway mechanisms
Sign up now to receive a 20% welcome discount
Institutional Access
How does access work?
Have an idea on how to improve our website?
Please write us.
© 2025 De Gruyter Brill
Downloaded on 8.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/med-2024-1110/html
Scroll to top button