Home Laboratory variables‐based artificial neural network models for predicting fatty liver disease: A retrospective study
Article Open Access

Laboratory variables‐based artificial neural network models for predicting fatty liver disease: A retrospective study

  • Panpan Lv , Zhen Cao , Zhengqi Zhu , Xiaoqin Xu and Zhen Zhao EMAIL logo
Published/Copyright: September 13, 2024
Download Article (PDF)
Open Medicine
From the journal Open Medicine Volume 19 Issue 1

Abstract

Background

The efficacy of artificial neural network (ANN) models employing laboratory variables for predicting fatty liver disease (FLD) remains inadequately established. The study aimed to develop ANN models to precisely predict FLD.

Methods

Of 12,058 participants undergoing the initial FLD screening, 7,990 eligible participants were included. A total of 6,309 participants were divided randomly into the training (4,415 participants, 70%) and validation (1,894 participants, 30%) sets for developing prediction models. The performance of ANNs was additionally tested in the testing set (1,681 participants). The area under the receiver operating characteristic curve (AUROC) was employed to assess the models’ performance.

Results

The 18-variable, 11-variable, 3-variable, and 2-variable models each achieved robust FLD prediction performance, with AUROCs over 0.92, 0.91, and 0.89 in the training, validation, and testing, respectively. Although slightly inferior to the other three models in performance (AUROC ranges: 0.89–0.92 vs 0.91–0.95), the 2-variable model showed 80.3% accuracy and 89.7% positive predictive value in the testing. Incorporating age and gender increased the AUROCs of the resulting 20-variable, 13-variable, 5-variable, and 4-variable models each to over 0.93, 0.92, and 0.91 in the training, validation, and testing, respectively.

Conclusions

Implementation of the ANN models could effectively predict FLD, with enhanced predictive performance via the inclusion of age and gender.

Keywords: fatty liver disease; artificial neural network; model; prediction; laboratory variables

1 Introduction

Fatty liver disease (FLD) is an increasingly prevalent global health issue, affecting over 25% of adults worldwide and posing a significant economic burden on society, the prediction and diagnosis of which is necessary for management and prognosis. However, early detection of FLD is challenging because of its silent and nonspecific symptomatology, compounded by limited technological capabilities for detection [1,2].

The current standard clinical workup for individuals suspected of or diagnosed with liver disease involves obtaining a comprehensive medical record, performing a thorough physical check, conducting laboratory tests, and interpreting imaging results [3]. Although these data modalities offer an abundance of information, their interpretation can be complex even for experienced clinicians. Hepatology is particularly prone to diagnostic ambiguities. Therefore, there is a need for advanced diagnostic approaches and improved technology to enhance the screening, early diagnosis, and subsequent intervention of FLD. Addressing these challenges is paramount alleviating the global burden of FLD and enhancing patient outcomes.

There have been several attempts to predict FLD. However, given its multifactorial nature, accurately predicting the occurrence of FLD using a single laboratory test parameter is unlikely. Ultrasonography (US) has been proposed as an initial screening modality for identifying steatosis in a specific cohort [4]. However, US has well-described limitations, particularly in its ability to detect focal liver lesions. These limitations pertain to a significant reliance on operator expertise, equipment standards, and patient physique [5]. Currently, liver biopsy is considered the diagnostic gold standard for assessing fatty infiltration of the liver and stratifying patients. Nonetheless, this invasive and costly method has its drawbacks, including the potential for side effects, sampling errors, and a lack of agreement among different observers [5]. Consequently, there is a growing demand for non-invasive or minor-invasive predictive models of FLD.

Studies are trying to find new markers or combined diagnoses for the early diagnosis of FLD to improve the sensitivity and clinical application. The construction of a prediction model to effectively identify high-risk groups and carry out targeted interventions is helpful not only for disease treatment but also for avoiding unnecessary excessive examinations and improving the utilization rate of medical resources.

Given the intricate nature of liver diseases and the often non-linear relationships between various variables and clinical outcomes, artificial neural network (ANN) has gained prominence in the past decade, particularly in medical model classification and assessment [3]. ANN is a robust machine learning model inspired by the neuroanatomy of the brain that is capable of non-linear statistical analysis. Comprising interconnected processing neurons with weighted connections, ANN forms a network structure that consists of an input layer, an output layer, and one or more hidden layers (Figure 1). Through training on extensive medical data, ANNs have the ability to extract hidden properties, offering a novel approach to effective discrimination [6]. In contrast, traditional statistical algorithms lack this adaptability, relying on explicit expressions of relationships [7]. In the case of diagnosing a specific condition, an ideal screening test should be simple to apply to the target population [8]. Therefore, prediction models of FLD based solely on laboratory test results plus demographic factors obtained at the time of examination using ANN were constructed in this study, so as to explore whether ANNs could serve as a promising strategy for FLD prediction from readily available tests. To validate this hypothesis, we evaluated and compared the predictive performance of eight different ANNs based on four distinct serum panels with and without the inclusion of demographic factors.

Figure 1 Data flowchart and architecture of the feed-forward ANN for the proposed ANN of this study. The ANN comprises an input layer, two hidden layers, and an output layer. Within each layer, a multitude of neurons, represented by solid circles, is present. The interconnections between these neurons are visually denoted by black lines.
Figure 1

Data flowchart and architecture of the feed-forward ANN for the proposed ANN of this study. The ANN comprises an input layer, two hidden layers, and an output layer. Within each layer, a multitude of neurons, represented by solid circles, is present. The interconnections between these neurons are visually denoted by black lines.

2 Materials and methods

2.1 Patients

Patients who had undergone initial screening for fatty liver examination at the physical examination center of Minhang Hospital, Fudan University (Shanghai, China) from 2021 to 2022 were enrolled in this study. Individuals with incomplete screening processes, known liver-related diseases, such as viral hepatitis (HCV, HBV), or suspected cases of fatty liver identified by US were excluded.

To construct and evaluate the performance of ANNs, the population between January 1, 2021, and July 31, 2022, included in this study was randomly divided into two subsets with 70% of patients (N = 4,415) in the training set and the remaining 30% (N = 1,894) in the validation set; 1,681 additional patients were enrolled between August 1, 2022, and December 31, 2022, to constitute the testing set, which was used to further assess the efficacy of the established models.

Blood tests were conducted on the day of the physical examination, following the detailed protocol outlined below. Two expert hepatologists used the previously published criteria [4,9,10] for the FLD diagnosis. In cases where there was a disagreement between the two hepatologists, a third experienced hepatologist was consulted to provide a judgment.

2.2 Clinical data acquisition

The clinical data used in our study were collected from both FLD and non-FLD patients, including 39 blood variables, which were determined at the time of screening and used in our proposed models. Thirty-nine blood variables (18 clinical chemistry variables and 21 complete blood counts) were as follows: (1) 18 clinical chemistry variables: total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), alanine aminotransferase (ALT), aspartate aminotransferase (AST), total protein (TP), total bilirubin (TBIL), direct bilirubin (DBIL), serum albumin (ALB), globulin (GLB), A/G, alkaline phosphatase (ALP), gamma-glutamyltransferase (GGT) level, lactic dehydrogenase (LDH), ureophil (URE), creatinine (CRE), and serum uric acid (UA), and (2) 21 complete blood counts: erythrocyte count, hemoglobin, neutrophil, lymphocytes, monocyte, acidophilic cell, basophilic granulocyte, neutrophil ratio%, ratio of lymphocytes%, monocyte%, acidophilic cell%, basophilic granulocyte%, monocyte%, peripheral platelet count, packed cell volume, mean corpuscular hemoglobin, mean corpuscular hemoglobin concentration, red cell distribution width SD, red cell distribution width CV, mean platelet volume, and platelet distribution width. All FLD and non-FLD patients were identified by the abdominal US. Clinical chemistry tests were performed using a Cobas c 702 Analyzer (Cobas, Germany), and the complete blood counts were measured on a BC-6000 Analyzer (Mindray, China). The prediction models with optimized variables were utilized to identify high-risk FLD patients, allowing for individualized health treatment in FLD patients.

2.3 Statistical analysis

Continuous variables were presented as the mean ± standard deviation. Comparisons between groups of quantitative variables were made by the one-way analysis of variance or Welch test, when appropriate. Statistical analysis was performed using the SPSS software version 20 (SPSS Inc., USA) and statistical significance was reported as p-values below 0.05.

2.4 Model development

The ANNs’ training and performance evaluation were conducted using Matlab with the neural pattern recognition application of the neural network toolbox from MathWorks, the Netherlands. Initially, we constructed models using only blood variables. A four-layer feed-forward neural network with a single output neuron was constructed (Figure 1). The backpropagation of errors learning rule was employed, allowing the network’s internal variables to be adjusted over repeated training cycles to minimize the overall error [11]. The activation function, representing the ANN’s outcomes, produced continuous outputs within the interval from 0 to 1 [12], where 0 indicated non-FLD and 1 represented FLD. Signal propagation occurred from the input layer through two hidden layers before reaching the output layer (Figure 1). These layers were fully connected, meaning that any neuron in the upper layer was connected to all neurons in the lower layer. The ANN training process involved randomly dividing the dataset into a training set (70% of total patients) for determining the network’s architecture and establishing the weights between nodes. A validation set (30% of total patients) was then utilized to evaluate the ANN’s ability to predict the desired output. Lastly, the neural network’s performance was assessed using an independent testing set. During training, the connection weights between neurons were adjusted iteratively to minimize the overall error. Training ceased when the sum of squared errors reached a minimum [12,13,14]. The number of neurons in the input layer was determined by the input data, with an n-dimensional vector included. Denoting the input layer as X, the output of the hidden layer was given by f(W 1 X + b 1), in which W 1 represented the weights, b 1 denoted the biases, and function f commonly employed activation functions (sigmoid function) [15]. Data propagation between the hidden layer and output layer followed softmax regression. The output of the output layer was softmax(W 2 X 1 + b 2), where X 2 (equal to f(W 1 X + b 1)) represented the hidden layer’s output. The formulation for the four-layer ANN described above can be summarized as follows [15]:

(1) f ( x ) = G ( b ( 2 ) + W ( 2 ) ( s ( b ( 1 ) + W ( 1 ) x ) ) ) .

The function G denotes the softmax function as described previously. Therefore, all variables of the ANN symbolize the connection weights and the bias between the layers, encompassing W 1, b 1, W 2, and b 2 [15].

2.5 Model evaluation

The model’s prediction performance was evaluated using a confusion matrix, area under the receiver operating characteristic curve (AUROC), and classification accuracy. The receiver operating characteristic curve (ROC) methodology, which is closely related to neural networks in classification applications, was employed [16]. The AUROC, a commonly used accuracy index, was calculated to assess the diagnostic accuracy, with values close to 1 indicating higher accuracy [8]. We assessed the weight of each variable by calculating the AUROC, respectively, to evaluate the prediction performance of the models. To test whereas ANNs, based on readily available and inexpensive variables, may add performance to the prediction of FLD, in the training group, four types of ANNs were developed for predicting FLD, that is, 18-variable model (model 1, including TC, TG, HDL-c, LDL-c, ALT, AST, TP, TBIL, DBIL, ALB, GLB, A/G, ALP, GGT, LDH, URE, CRE, and UA), 11-variable model (model 2, including ALT, AST, TG, TC, r-GT, LDL-c, HDL-c, ALP, LDH, CRE, and UA), 3-variable model (model 3, including ALT, TG, and HDL), and 2-variable model (model 4, including ALT and TG). These four models were selected based on their specific characteristics. Firstly, all 39 blood variables were used for training and the best subset of relevant parameters was identified for subsequent model building. The automatic classification process registered approximately 75% sensitivity for FLD and non-FLD, which demonstrated that the classification performance was poor. Because of the overall small effect on the classification performance, the 21 routine blood variables were abandoned. After subtracting of routine blood indicators, 18 serum variables were trained (18-variable model). However, some variables are not readily available, especially for basic hospitals, and are expensive in routine diagnosis and treatment. To select the most promising predictive variables and achieve the highest predictive accuracy for FLD prediction, we retained variables with AUROC greater than 0.6 as input for the model. It helped to assess the efficacy of the variables incorporated in the training set. Thereafter, any invalid variables for classification were eliminated, resulting in an 11-variable model. Finally, to identify inexpensive and readily available variables with the least amount of detection requirements to aid clinicians in predicting FLD, ensuring patients receive appropriate and accurate treatment while maximizing the utilization of medical resources, we traversed three variables and two variables from the 18 variables to determine the optimal combination. We systematically evaluated each combination’s AUROC of any 3-variable combinations, which constitute 816 possible combinations (C (18,3) = 18 × 17 × 16/(3!) = 816), and any 2-variable combinations, totaling 153 possible combinations (C (18,2) = 18 × 17/(2!) = 153). Subsequently, we selected the combination with the highest AUROC as the optimal configuration for both the 3-variable and 2-variable models. It helped in evaluating the effectiveness. This meticulous process enabled us to derive the performance metrics for both models based on their respective optimal combinations of variables. Consequently, we developed both a 3-variable model and a 2-variable model. The final weights of the variables for the 2-variable and 3-variable models were determined based on the AUROC of all the training, validation, and testing sets. We also calculated the sensitivity (SEN), specificity (SPE), accuracy (ACC), false positive rate (FPR), and positive predictive value (PPV). Moreover, to further analyze the potential value of incorporating additional easily collectible and demographic factors, such as age and gender, into the constructed models, we incorporated these two factors for advanced analyses, resulting in a 20-variable model (model 5, including the variables of 18-variable model plus age and gender), a 13-variable model (model 6, including the variables of 11-variable model plus age and gender), a 5-variable model (model 7, including the variables of 3-variable model plus age and gender), and a 4-variable model (model 8, including the variables of 2-variable model plus age and gender).

The confusion matrix (Table 1) was used to determine the relationship between the actual values and predicted values [17]. The horizontal and vertical coordinates of the receiver operating characteristic curve are represented using 1 − specificity and sensitivity, respectively:

(2) Sensitivity ( SEN ) = TP / ( TP + FN ) ,

(3) Specificity ( SPE ) = TN / ( TN + FP ) ,

Table 1

Confusion matrix representation

Reality Positive Negative
Predicted true (+) TP (true positive) TN (true negative)
Predicted false (−) FP (false positive) FN (false negative)

Other evaluation indexes are calculated as follows:

(4) Accuracy ( ACC ) = ( TP + TN ) / ( TP + FP + TN + FN ) ,

(5) Positive predictive value ( PPV ) = TP / ( TP + TN ) .

In light of the aforementioned evaluation indexes of the model, we introduced the indicators in the optimal FLD prediction model, by comparing the results of different modeling methods to determine the optimal modeling method for the research data.

  1. Ethics approval: The research has been complied with all the relevant national regulations, institutional policies and in accordance with the tenets of the Helsinki Declaration, and has been approved by the Ethics Committee of Minhang Hospital, Fudan University (047-01K).

  2. Informed consent: The review board of the Ethics Committee deemed the study exempt from review and waived the requirement for informed consent due to the utilization of only de-identified data.

3 Results

3.1 Patient characteristics

A total of 12,058 participants who underwent initial fatty liver screening at Minghang Hospital, Fudan University between January 1, 2021, and December 31, 2022, were identified. Among them, 4,068 patients were excluded due to incomplete examination, resulting in a final sample size of 7,990 patients who fulfilled all inclusion criteria for model development (Figure 2). Of the overall subjects, 4,495 (56.3%) were patients with non-FLD, and the remaining 3,495 (43.7%) were FLD. Their mean age was 37.3 ± 8.5 years; 3,847 (48.1%) were male and 4,143 (51.9%) were female. A comparison between patients with and without FLD is presented in Table 2, revealing all variables (except for DBIL) to be statistically significant factors associated with FLD. Additional details regarding the distribution of variables of the FLD and non-FLD groups within the training, validation, and testing sets, are shown in Table 3. The training set comprised 4,415 subjects (1,717 FLD and 2,698 non-FLD), the validation set included 1,894 subjects (737 FLD and 1,157 non-FLD), and the testing set contained 1,681 subjects (1,041 FLD and 640 non-FLD). In the training set, all variables (except for DBIL) were found to be statistically significant factors associated with FLD. Variables (except for DBIL, A/G, and GLB) were found to be statistically significant factors associated with FLD in the validation set. Whereas, in the testing set, all variables were found to be statistically significant factors associated with FLD (Table 3).

Figure 2 Workflow for patient screening.
Figure 2

Workflow for patient screening.

Table 2

Comparing characteristics of patients with and without significant FLD

Variables Reference range FLD (N = 3,495) Non-FLD (N = 4,495) p-value
Male gender 2,615(74.8%) 1,231(27.4%) <0.001
Median age (years) 39.9(±8.5) 35.3(±7.9) <0.001
TC (mmol/L) 2.8–5.9 4.9(±0.9) 4.5(±0.8) <0.001
TG (mmol/L) 0–2.3 2.4(±2.4) 1.1(±0.6) <0.001
HDL (mmol/L) 0.9–1.68 1.1±(0.3) 1.6(±0.4) <0.001
LDL (mmol/L) <3.1 3.3(±0.8) 2.8(±0.7) <0.001
ALT (U/L) 0–66 38.2(±29.5) 15.2(±13.7) <0.001
AST (U/L) 0–40 24.7(±13.8) 17.5(±9.3) <0.001
TP (g/L) 64–83 75.8(±3.8) 74.8(±3.9) <0.001
TBIL (μmol/L) 2–20 11.9(±5.4) 11.1(±5.3) <0.001
DBIL (μmol/L) 0–6 4.4(±1.6) 4.3(±1.6) 0.091
A/G 1.1–1.8 1.8(±0.3) 1.9(±0.3) 0.045
GLB (g/L) 29–33 27.4(±3.6) 26.6(±3.5) <0.001
ALB (g/L) 35–50 48.4(±2.6) 48.2(±2.7) 0.001
ALP (U/L) 39–120 77.4(±19.9) 64.1(±22.1) <0.001
GGT (U/L) 0–54 46.5(±41.5) 18.8(±19.4) <0.001
LDH (U/L) 135–225 169.2(±28.5) 155.9(±23.9) <0.001
URE (mmol/L) 1.7–8.3 4.8(±1.2) 4.6(±1.2) <0.001
CRE (μmol/L) 20–110 78.3(±18.4) 67.6(±1.9) <0.001
UA (μmol/L) 142–416 397.0(±90.1) 294.8(±77.6) <0.001

Results are shown as the mean ± SD. p-values were calculated between the data of the FLD and the non-FLD. Statistical significance was reported as p-values below 0.05.

Table 3

Comparing characteristics of patients with and without significant FLD within a set (i.e., training, validation, or testing)

Training set (N = 4,415) Validation set (N = 1,894) Testing set (N = 1,681)
Variables Reference range FLD (N = 1,717) Non-FLD (N = 2,698) p 1-value FLD (N = 737) Non-FLD (N = 1,157) p 2-value FLD (N = 1,041) Non-FLD (N = 640) p 3-value
Male gender 1,280(74.5%) 798(29.6%) <0.001 570(77.3%) 335(29.0%) <0.001 765(73.5%) 98(15.3%) <0.001
Median age (years) 39.0(±7.1) 35.4(±7.9) <0.001 39.3(±7.3) 34.9(±7.9) <0.001 41.7(±11.0) 35.7(±8.3) <0.001
TC (mmol/L) 2.8–5.9 4.9(±0.9) 4.4(±0.8) <0.001 4.8(±0.8) 4.4(±0.7) <0.001 5.1(±0.9) 4.8(±0.9) <0.001
TG (mmol/L) 0–2.3 2.5(±2.8) 1.1(±0.6) <0.001 2.4(±1.8) 1.0(±0.6) <0.001 2.3(±2.0) 1.0(±0.6) <0.001
HDL (mmol/L) 0.9–1.68 1.1(±0.2) 1.5(±0.4) <0.001 1.1(±0.3) 1.5(±0.3) <0.001 1.2(±0.3) 1.6(±0.4) <0.001
LDL (mmol/L) <3.1 3.3(±0.8) 2.8(±0.7) <0.001 3.2(±0.8) 2.8(±0.7) <0.001 3.2(±0.8) 2.9(±0.8) <0.001
ALT (U/L) 0–66 39.1(±30.6) 15.4(±15.2) <0.001 37.5(±28.5) 14.6(±10.6) <0.001 37.5(±28.5) 15.8(±11.7) <0.001
AST (U/L) 0–40 25.6(±15.4) 17.8(±10.1) <0.001 24.7(±12.8) 17.6(±8.8) <0.001 23.4(±11.2) 16.3(±6.2) <0.001
TP (g/L) 64–83 75.6(±3.8) 74.8(±3.9) <0.001 75.5(±3.8) 75.0(±3.8) 0.005 76.2(±3.8) 74.4(±3.9) <0.001
TBIL (μmol/L) 2-20 11.7(±5.3) 11.2(±5.4) 0.007 12.2(±5.9) 11.1(±5.2) <0.001 12.0(±5.2) 10.6(±4.8) <0.001
DBIL (μmol/L) 0–6 4.3(±1.5) 4.3(±1.6) 0.332 4.4(±1.6) 4.3(±1.6) 0.184 4.5(±1.6) 4.2(±1.5) 0.001
A/G 1.1–1.8 1.8(±0.3) 1.9(±0.3) 0.02 1.9(±0.3) 1.9(±0.3) 0.869 1.7(±0.3) 1.8(±0.3) <0.001
GLB (g/L) 29–33 26.9(±3.5) 26.4(±3.5) <0.001 26.8(±3.4) 26.6(±3.3) 0.217 28.6(±3.6) 27.4(±3.6) <0.001
ALB (g/L) 35–50 48.7(±2.6) 48.3(±2.6) <0.001 48.7(±2.6) 48.4(±2.7) 0.014 47.6(±2.5) 47.0(±2.5) <0.001
ALP (U/L) 39–120 76.6(±19.2) 63.9(±21.9) <0.001 77.6(±21.1) 64.7(±24.6) <0.001 78.6(±20.2) 63.4(±17.2) <0.001
GGT (U/L) 0–54 46.6(±44.3) 19.0(±20.9) <0.001 46.9(±39.0) 18.0(±16.2) <0.001 46.2(±38.3) 19.4(±17.7) <0.001
LDH (U/L) 135–225 167.3(±27.6) 155.5(±23.6) <0.001 168.1(±28.1) 156.0(±24.5) <0.001 173.0(±30.0) 157.4(±23.5) <0.001
URE (mmol/L) 1.7–8.3 4.8(±1.3) 4.6(±1.1) <0.001 4.9(±1.1) 4.6(±1.2) <0.001 4.7(±1.1) 4.5(±1.1) <0.001
CRE (μmol/L) 20–110 78.2(±21.4) 68.1(±14.0) <0.001 79.1(±15.2) 67.9(±14.1) <0.001 77.8(±14.5) 65.1(±12.6) <0.001
UA (μmol/L) 142–416 397.5(±90.8) 295.9(±76.7) <0.001 397.8(±88.7) 296.3(±80.5) <0.001 395.5(±89.9) 287.8(±75.8) <0.001

Results are shown as the mean ± SD. The training set comprised 4,415 subjects (1,717 FLD and 2,698 non-FLD), the validation set included 1,894 subjects (737 FLD and 1,157 non-FLD), and the testing set contained 1,681 subjects (1,041 FLD and 640 non-FLD). p 1, p 2, and p 3 were calculated among the data of patients with and without significant FLD within the training, validation, and testing sets, respectively. Statistical significance was reported as p-values below 0.05.

4 Model performance

4.1 Training group

Through the training on the training set, the four models – 18-variable, 11-variable, 3-variable, and 2-variable, all achieved a very robust performance on FLD prediction, with AUROCs of 0.95, 0.94, 0.93, and 0.92, respectively. When age and gender were incorporated, the AUROC for the four models – 20-variable, 13-variable, 5-variable, and 4-variable, slightly increased, resulting in 0.95, 0.95, 0.94, and 0.93, respectively. Detailed quantitative results of AUROC, ACC, SEN, SPE, FPR, and PPV for each model are presented in Table 4. Notably, with a cut-off value of 0.5, models 1–4 had an excellent accuracy of over 85%, a sensitivity of over 80%, and a specificity of over 87%. Similarly, models 5–8 maintained excellent accuracy over 85%, sensitivity over 80%, and specificity over 88%. The performance metrics of the 2-variable model and 4-variable model were universally comparable to those of any other models. The ROC curves of models 1–4 and models 5–8 in the training set are shown in Figure 3(a) and (d).

Table 4

Performance of models for differentiation of patient groups of FLD by ANN

Parameter Model 1 (18 variables) Model 2 (11 variables) Model 3 (3 variables) Model 4 (2 variables) Model 5 (20 variables) Model 6 (13 variables) Model 7 (5 variables) Model 8 (4 variables)
Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set Training set Validation set Testing set
Threshold 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5
AUROC 0.95 0.92 0.91 0.94 0.93 0.91 0.93 0.92 0.91 0.92 0.91 0.89 0.95 0.93 0.92 0.94 0.94 0.92 0.94 0.92 0.92 0.93 0.92 0.91
SEN 84.7% 81.3% 83.3% 86.2% 85.6% 82.9% 82.4% 82.0% 78.2% 80.0% 80.6% 77.0% 85.0% 83.5% 87.4% 82.9% 79.1% 83.5% 82.6% 81.1% 83.0% 80.7% 79.2% 79.8%
SPE 90.1% 89.3% 85.5% 87.6% 86.3% 83.8% 87.3% 88.1% 87.0% 88.1% 85.6% 85.6% 89.5% 89.0% 84.2% 89.7% 90.7% 87.0% 89.0% 87.7% 85.8% 88.7% 88.6% 86.4%
ACC 88.0% 86.2% 84.1% 87.1% 86.1% 83.2% 85.4% 85.6% 81.6% 85.1% 83.6% 80.3% 87.8% 86.9% 86.2% 87.0% 86.2% 84.8% 86.5% 85.2% 84.1% 85.6% 85.0% 82.3%
FPR 9.9% 10.7% 14.5% 12.4% 13.7% 16.3% 12.8% 11.9% 13.0% 11.9% 14.4% 14.4% 10.5% 11.0% 15.8% 10.3% 9.3% 13.0% 11.0% 12.3% 14.2% 11.3% 11.4% 13.6%
PPV 84.5% 82.8% 90.3% 81.6% 80.0% 89.3% 80.4% 81.4% 90.8% 81.1% 78.1% 89.7% 83.8% 82.9% 90.0% 83.7% 84.4% 91.3% 82.7% 80.8% 90.5% 82.0% 81.6% 90.5%

Abbreviations: AUROC, area under the receiver operating characteristic curve; ACC, accuracy; SEN, sensitivity; SPE, specificity; FPR, false positive rate; PPV, positive predictive value. There were 1,717 (training set), 737 (validation set), and 1,041 (testing set) patients with FLD and 2,698 (training set), 1,157 (validation set), and 640 (testing set) patients with non-FLD.

Figure 3 The ROC curves of the eight different models for the prediction of FLD in the training set (a and d), the validation set (b and e), and the testing set (c and f).
Figure 3

The ROC curves of the eight different models for the prediction of FLD in the training set (a and d), the validation set (b and e), and the testing set (c and f).

4.2 Validation group

When the ANNs were evaluated in the validation group, the models in predicting FLD yielded AUROCs of 0.92, 0.93, 0.92, and 0.91 among models 1–4, while models 5–8 yielded AUROCs of 0.93, 0.94, 0.92, and 0.92. Although the AUROC values were universally slightly lower than those obtained in the training group, they were still regarded as strong indicators of model performance. The quantitative results corresponding to the validation are summarized in Table 4. The ROC curves yielded by the models in the validation set are plotted in Figure 3(b) and (e) for the first four and last four models.

4.3 Testing group

The performance provided by the 18-variable model, the 11-variable model, and the 3-variable model in the testing group had an overall high predictive ability with AUROC of 0.91. Although the 2-variable model’s performance was slightly lower than that of the other three models (0.89 vs 0.91), it showed a comparable performance among them (Table 4). When age and gender were added into the models, the resulting 20-variable, 13-variable, 5-variable, and 4-variable models exhibited AUROCs of over 0.91 in the testing group. This slight improvement in AUROC values highlights the added predictive value of including age and gender. In particular, despite the TBIL, DBIL, URE, and CRE, the other serum variables in the testing set were significantly different from those in the training set and the validation set (Table S1), which demonstrated the well generalization ability of the ANNs in turn. Especially, it was indicated that the proposed ANNs exhibited high predictive ability in FLD screening. The ROC curves of these methods in the testing set are presented in Figure 3(c) and (f) for the first four and last four models.

5 Discussion

FLD has become a significant global health concern due to its increasing prevalence and widespread occurrence across different age groups. In recent years, the utilization of machine learning models, particularly ANNs, has presented a unique opportunity to enhance the comprehensive management of FLD. These models have demonstrated their potential in improving the risk prediction and diagnosis of FLD [18]. By analyzing medical variables, machine learning models, such as ANNs, offer an efficient approach to uncover hidden relationships among variables that may otherwise go unnoticed. Their ability to extract hidden associations from complex and diverse clinical datasets has been well established [5].

The confusion matrix analysis showed that the ANNs using serum variables could achieve their predictive purpose. In the present study, we compared the combination of the serum-based panels as predictors, respectively. Eventually, the 18-variable model, the 11-variable model, the 3-variable model, and the 2-variable model were developed, all having the same two core predictors, ALT and TG. Whereas, compared to the 2-variable model, the 3-variable model included an extra predictor, HDL. What is more, to assess the increased value of demographic factors, we incorporated age and gender into our existing models. This resulted in a 20-variable model, a 13-variable model, a 5-variable model, and a 4-variable model.

ANNs utilizing serum variables have exhibited promising predictive capabilities in the field of FLD. The evaluation of ROC curves has demonstrated high predictive accuracies for models 1–4 in the training, validation, and testing sets. The AUROCs were 0.91, 0.91, 0.91, and 0.89 among the testing sets of the first four models, respectively, which further confirmed the robustness of ANNs employing serum variables as a reliable method for FLD prediction. Notably, the primary variables influencing FLD prediction were observed to be ALT and TG. The 2-variable model exhibited comparable performance to the 18-variable, 11-variable, and 3-variable models, indicating that an effective distinction between FLD and non-FLD patients can be achieved through a simplified approach.

Interestingly, when age and gender were included in the first four models, it resulted in AUROCs of 0.92, 0.92, 0.92, and 0.91 for the 20-variable, 13-variable, 5-variable, and 4-variable models among the testing sets, respectively. The slight improvement in AUROCs highlights the enhanced predictive performance with the inclusion of age and gender, indicating that these two factors are critical elements in effectively distinguishing between FLD and non-FLD patients. Despite this enhancement, the 4-variable model with age and gender demonstrated comparable performance to more complex models, reinforcing its potential as a superior option for accurately and effectively predicting FLD. Consequently, our findings suggest that the implementation of an uncomplicated model could serve as a superior option for accurately and effectively predicting FLD. These results warrant consideration for the development of an appropriate predictive system in this domain.

Attempts have been made to enhance the performance of FLD prediction by incorporating additional variables through machine learning techniques. As early as 2014, Vanderbeck et al. [19] employed a support vector machine algorithm with handcrafted features to identify and quantify various structures on scanned Hematoxylin and Eosin slides from nonalcoholic FLD (NAFLD ) and healthy liver biopsies, achieving an overall accuracy of 89%. Roy et al. [20] developed a U-Net architecture algorithm that effectively segmented and quantified hepatic steatosis. Lin et al. [21] utilized multivariate analysis incorporating sex, age, TG, BMI, TC, and ALT, indicating that multinomial logistic regression (LR) exhibited the highest predictive power, with an accuracy rate of 72.6% for first-degree FLD and 62.3% for second- and third-degree FLD. Islam et al. [22] developed four classification models – Random Forest (RF), Support Vector Machine (SVM), ANN, and LR for FLD prediction, with LR yielding the best results (76.3% accuracy, 74.1% sensitivity, 64.9% specificity). In addition, Wu et al. [23] created four classification models – RF, Naïve Bayes (NB), ANN, and LR – to evaluate the optimal predictive clinical model for FLD, where the RF exhibited superior performance with 10-fold cross-validation, achieving an accuracy of 86.48% and an AUROC of 0.925. This model incorporated 10 clinical values, including age, gender, systolic blood pressure, diastolic blood pressure, abdominal girth, glucose AC, TG, HDL-C, AST, and ALT. Okanoue et al. [18] reported that the artificial intelligence/neural network system utilizing 11 medical values (including age, gender, height, weight, waist circumference, AST, ALT, GGT, cholesterol, TG, and PLT) had well efficacy in diagnosing NAFLD (AUROC > 0.950). Overall, these studies showed the potential of machine learning technology for identification in patients with FLD.

Although the application and assessment of machine learning had been explored for the recognition of FLD using tongue images [24], liver biopsy images [19], US [25,26], clinical data [17,27,28], and a combination of US and clinical data [17,27,29], a promising model for FLD prediction only on the basis of serum data with very few variables has seldom been applied in routine clinical care. Even though differences in variables were observed between patients with and without FLD, the discrepancies in test outcomes were not significant enough to utilize a single biomarker as an independent predictor of FLD. Therefore, we employed ANN to integrate only serum variables alone and in combination with age and gender that accurately classify patients at high risk of FLD during examination. Particularly, the 2-variable model using TG and ALT and the 4-variable model, which incorporated age and gender into the 2-variable model, were found to be with good performance in the training set, the validation set, and the testing set. This fully demonstrated the superiority of the neural network and the well generalization ability of the ANNs, which were with sufficient accuracy to be usefully employed as a reliable and user-friendly tool for identifying FLD.

However, it is important to acknowledge the limitations of this study. One possible limitation is the choice of modeling methods, which can significantly impact the accuracy of disease prediction models. Therefore, future research aims to develop more advanced neural network models combined with image analysis to facilitate the diagnosis of FLD. Additionally, the data collection was limited to one medical center. There may be selection bias in the data selection process, which needs to be handled carefully to ensure the reliability and validity of the research results. Multicenter datasets should be sought to further improve the reliability and clinical usability of the constructed models. Further research in this area, exploring the utilization of ANN and other machine learning technologies, holds promise for improved results and enhanced preventive healthcare.

6 Conclusion

In conclusion, this study successfully developed an ANN-based variables integration model by constructing different FLD prediction models, to integrate the information of only serum variables alone and alongside age and gender for accurately predicting patients with the FLD, allowing patients to be tested for just two serum markers (ALT and TG) to determine if further diagnostic tests are needed, which avoids unnecessary and excessive medical treatment, so that patients can receive appropriate treatment at an early stage. We confirmed the application value of the prediction of FLD, providing strong support for the follow-up application in disease prediction. The FLD prediction model combined with serum variables having well repeatability, reproducibility, and generalization ability, is worthy of further exploration. We anticipate the functionality of the system to provide significant patient benefits.

Abbreviations

ACC

Accuracy

ANN

Artificial neural network

AUROC

Area under the receiver operating characteristic curve

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

ALB

Albumin

ALP

Alkaline phosphatase

CRE

Creatinine

DBIL

Direct bilirubin

FLD

Fatty liver disease

FPR

False positive rate

GLB

Globulin

GGT

Gamma-glutamyltransferase

HDL-c

High-density lipoprotein cholesterol

LDH

Lactic dehydrogenase

LDL-c

Low-density lipoprotein cholesterol

LR

Logistic regression

NB

Naïve Bayes

NAFLD

Nonalcoholic fatty liver disease

PPV

Positive predictive value

RF

Random Forest

SVM

Support Vector Machine

SEN

Sensitivity

SPE

Specificity

TC

Total cholesterol

TG

Triglyceride

TP

Total protein

TBIL

Total bilirubin

UA

Uric acid

URE

Ureophil

US

Ultrasound

Acknowledgements

The authors thank Mindray Medical International Limited, China, for their assistance in methodology.

  1. Funding information: This work was supported by the training program for outstanding young medical talents and pharmaceutical talents of Shanghai Minhang District Health Commission (mwyjyx08).

  2. Author contributions: PPL and ZZ contributed to study design, data collection, funding acquisition, and manuscript preparation. ZC, ZQZ, and XQX were responsible for statistical analysis. All authors read and approved the final manuscript.

  3. Conflict of interest: The authors state that there are no conflicts of interest to disclose. Mindray Medical International Limited, China, states that they do not have any relevant competing interests, including financial and otherwise, to disclose.

  4. Code availability statement: The code used in the study was developed by Mindray Medical International Limited, China. Due to the stringent information security regulations mandated by the company, we are unable to provide access to the code. The code contains proprietary information owned by Mindray, and as such, we are bound by strict confidentiality protocols preventing its export.

  5. Data availability statement: The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

References

[1] Riazi K, Azhari H, Charette JH, Underwood FE, King JA, Afshar EE, et al. The prevalence and incidence of NAFLD worldwide: a systematic review and meta-analysis. Lancet Gastroenterol Hepatol. 2022;7(9):851–61.10.1016/S2468-1253(22)00165-0Search in Google Scholar PubMed

[2] Keles U, Ow JR, Kuentzel KB, Zhao LN, Kaldis P. Liver-derived metabolites as signaling molecules in fatty liver disease. Cell Mol Life Sci. 2022;80(1):4.10.1007/s00018-022-04658-8Search in Google Scholar PubMed PubMed Central

[3] Nam D, Chapiro J, Paradis V, Seraphin TP, Kather JN. Artificial intelligence in liver diseases: Improving diagnostics, prognostics and response prediction. JHEP Rep. 2022;4(4):100443.10.1016/j.jhepr.2022.100443Search in Google Scholar PubMed PubMed Central

[4] Papatheodoridi M, Cholongitas E. Diagnosis of non-alcoholic fatty liver disease (NAFLD): current concepts. Curr Pharm Des. 2018;24(38):4574–86.10.2174/1381612825666190117102111Search in Google Scholar PubMed

[5] Calderaro J, Seraphin TP, Luedde T, Simon TG. Artificial intelligence for the prevention and clinical management of hepatocellular carcinoma. J Hepatol. 2022;76(6):1348–61.10.1016/j.jhep.2022.01.014Search in Google Scholar PubMed PubMed Central

[6] Le Berre C, Sandborn WJ, Aridhi S, Devignes MD, Fournier L, Smaïl-Tabbone M, et al. Application of artificial intelligence to gastroenterology and hepatology. Gastroenterology. 2020;158(1):76–94.e2.10.1053/j.gastro.2019.08.058Search in Google Scholar PubMed

[7] Reibnegger G, Weiss G, Werner-Felmayer G, Judmaier G, Wachter H. Neural networks as a tool for utilizing laboratory information: comparison with linear discriminant analysis and with classification and regression trees. Proc Natl Acad Sci USA. 1991;88(24):11426–30.10.1073/pnas.88.24.11426Search in Google Scholar PubMed PubMed Central

[8] Procopet B, Cristea VM, Robic MA, Grigorescu M, Agachi PS, Metivier S, et al. Serum tests, liver stiffness and artificial neural networks for diagnosing cirrhosis and portal hypertension. Dig Liver Dis. 2015;47(5):411–6.10.1016/j.dld.2015.02.001Search in Google Scholar PubMed

[9] European Association for the Study of the Liver (EASL); European Association for the Study of Diabetes (EASD); European Association for the Study of Obesity (EASO). EASL-EASD-EASO Clinical practice guidelines for the management of non-alcoholic fatty liver disease. J Hepatol. 2016;64(6):1388–2.10.1016/j.jhep.2015.11.004Search in Google Scholar PubMed

[10] Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, et al. The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases. Hepatology. 2018;67(1):328–7.10.1002/hep.29367Search in Google Scholar PubMed

[11] Rumelhart DE, Hinton GE, Williams RJ. Learning representations by back-propagating errors. Nature. 1986;323:533–6.10.1038/323533a0Search in Google Scholar

[12] Zheng MH, Shi KQ, Lin XF, Xiao DD, Chen LL, Liu WY, et al. A model to predict 3-month mortality risk of acute-on-chronic hepatitis B liver failure using artificial neural network. J Viral Hepat. 2013;20(4):248–5.10.1111/j.1365-2893.2012.01647.xSearch in Google Scholar PubMed

[13] Baxt WG. Application of artificial neural networks to clinical medicine. Lancet. 1995;346(8983):1135–8.10.1016/S0140-6736(95)91804-3Search in Google Scholar

[14] Dayhoff JE, DeLeo JM. Artificial neural networks: opening the black box. Cancer. 2001;91(8 Suppl):1615–35.Search in Google Scholar

[15] Fan Z, Guo Y, Gu X, Huang R, Miao W. Development and validation of an artifcial neural network model for non-invasive gastric cancer screening and diagnosis. Sci Rep. 2022;12:21795.10.1038/s41598-022-26477-4Search in Google Scholar PubMed PubMed Central

[16] Dayhoff JE, Deleo JM. Artificial neural network-Opening the black box. Cancer. 2001;91:1616–35.10.1002/1097-0142(20010415)91:8+<1615::AID-CNCR1175>3.0.CO;2-LSearch in Google Scholar

[17] Perveen S, Shahbaz M, Keshavjee K, Guergachi A. Systematic machine learning based approach for the diagnosis of non-alcoholic fatty liver disease risk and progression. Sci Rep. 2018;8:2112.10.1038/s41598-018-20166-xSearch in Google Scholar

[18] Okanoue T, Shima T, Mitsumoto Y, Umemura A, Yamaguchi K, Itoh Y, et al. Artificial intelligence/neural network system for the screening of nonalcoholic fatty liver disease and nonalcoholic steatohepatitis. Hepatol Res. 2021;51(5):554–9.10.1111/hepr.13628Search in Google Scholar

[19] Vanderbeck S, Bockhorst J, Komorowski R, Kleiner DE, Gawrieh S. Automatic classifification of white regions in liver biopsies by supervised machine learning. Hum Pathol. 2014;45:785–2.10.1016/j.humpath.2013.11.011Search in Google Scholar

[20] Roy M, Wang F, Vo H, Teng D, Teodoro G, Farris AB, et al. Deep-learning based accurate hepatic steatosis quantifification for histological assessment of liver biopsies. Lab Invest. 2020;100:1367–83.10.1038/s41374-020-0463-ySearch in Google Scholar

[21] Lin YC, Chou SC, Huang PT, Chiou HY. Risk factors and predictors of non-alcoholic fatty liver disease in Taiwan. Ann Hepatol. 2011;10(2):125–2.10.1016/S1665-2681(19)31560-1Search in Google Scholar

[22] Islam MM, Wu CC, Poly TN, Yang HC, Li YC. Applications of Machine Learning in Fatty Live Disease Prediction: 40th Medical Informatics in Europe Conference, MIE 2018. IOS Press; 2018. p. 166–70.Search in Google Scholar

[23] Wu CC, Yeh WC, Hsu WD, Islam MM, Nguyen PAA, Poly TN, et al. Prediction of fatty liver disease using machine learning algorithms. Comput Methods Programs Biomed. 2019;170:23–9.10.1016/j.cmpb.2018.12.032Search in Google Scholar PubMed

[24] Jiang T, Guo XJ, Tu LP, Lu Z, Cui J, Ma XX, et al. Application of computer tongue image analysis technology in the diagnosis of NAFLD. Comput Biol Med. 2021;135:104622.10.1016/j.compbiomed.2021.104622Search in Google Scholar PubMed

[25] Kuppili V, Biswas M, Sreekumar A, Suri HS, Saba L, Edla DR, et al. Author correction to: extreme learning machine framework for risk stratification of fatty liver disease using ultrasound tissue characterization. J Med Syst. 2017;42:18.10.1007/s10916-017-0862-9Search in Google Scholar PubMed

[26] Byra M, Styczynski G, Szmigielski C, Kalinowski P, Michałowski Ł, Paluszkiewicz R, et al. Transfer learning with deep convolutional neural network for liver steatosis assessment in ultrasound images. Int J Comput Assist Radiol Surg. 2018;13:1895–3.10.1007/s11548-018-1843-2Search in Google Scholar PubMed PubMed Central

[27] Yip TC, Ma AJ, Wong VW, Tse YK, Chan HL, Yuen PC, et al. Laboratory parameter based machine learning model for excluding non-alcoholic fatty liver disease (NAFLD) in the general population. Aliment Pharmacol Ther. 2017;46:447–6.10.1111/apt.14172Search in Google Scholar PubMed

[28] Ma H, Xu CF, Shen Z, Yu CH, Li YM. Application of machine learning techniques for clinical predictive modeling: a cross-sectional study on nonalcoholic fatty liver disease in China. Biomd Res Int. 2018;2018:4304376.10.1155/2018/4304376Search in Google Scholar PubMed PubMed Central

[29] Spann A, Yasodhara A, Kang J, Watt K, Wang B, Goldenberg A, et al. Applying machine learning in liver disease and transplantation: a comprehensive review. Hepatology 2020;71:1093–5.10.1002/hep.31103Search in Google Scholar PubMed

Received: 2024-01-12
Revised: 2024-08-07
Accepted: 2024-08-12
Published Online: 2024-09-13

© 2024 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. EDNRB inhibits the growth and migration of prostate cancer cells by activating the cGMP-PKG pathway
  3. STK11 (LKB1) mutation suppresses ferroptosis in lung adenocarcinoma by facilitating monounsaturated fatty acid synthesis
  4. Association of SOX6 gene polymorphisms with Kashin-Beck disease risk in the Chinese Han population
  5. The pyroptosis-related signature predicts prognosis and influences the tumor immune microenvironment in dedifferentiated liposarcoma
  6. METTL3 attenuates ferroptosis sensitivity in lung cancer via modulating TFRC
  7. Identification and validation of molecular subtypes and prognostic signature for stage I and stage II gastric cancer based on neutrophil extracellular traps
  8. Novel lumbar plexus block versus femoral nerve block for analgesia and motor recovery after total knee arthroplasty
  9. Correlation between ABCB1 and OLIG2 polymorphisms and the severity and prognosis of patients with cerebral infarction
  10. Study on the radiotherapy effect and serum neutral granulocyte lymphocyte ratio and inflammatory factor expression of nasopharyngeal carcinoma
  11. Transcriptome analysis of effects of Tecrl deficiency on cardiometabolic and calcium regulation in cardiac tissue
  12. Aflatoxin B1 induces infertility, fetal deformities, and potential therapies
  13. Serum levels of HMW adiponectin and its receptors are associated with cytokine levels and clinical characteristics in chronic obstructive pulmonary disease
  14. METTL3-mediated methylation of CYP2C19 mRNA may aggravate clopidogrel resistance in ischemic stroke patients
  15. Understand how machine learning impact lung cancer research from 2010 to 2021: A bibliometric analysis
  16. Pressure ulcers in German hospitals: Analysis of reimbursement and length of stay
  17. Metformin plus L-carnitine enhances brown/beige adipose tissue activity via Nrf2/HO-1 signaling to reduce lipid accumulation and inflammation in murine obesity
  18. Downregulation of carbonic anhydrase IX expression in mouse xenograft nasopharyngeal carcinoma model via doxorubicin nanobubble combined with ultrasound
  19. Feasibility of 3-dimensional printed models in simulated training and teaching of transcatheter aortic valve replacement
  20. miR-335-3p improves type II diabetes mellitus by IGF-1 regulating macrophage polarization
  21. The analyses of human MCPH1 DNA repair machinery and genetic variations
  22. Activation of Piezo1 increases the sensitivity of breast cancer to hyperthermia therapy
  23. Comprehensive analysis based on the disulfidptosis-related genes identifies hub genes and immune infiltration for pancreatic adenocarcinoma
  24. Changes of serum CA125 and PGE2 before and after high-intensity focused ultrasound combined with GnRH-a in treatment of patients with adenomyosis
  25. The clinical value of the hepatic venous pressure gradient in patients undergoing hepatic resection for hepatocellular carcinoma with or without liver cirrhosis
  26. Development and validation of a novel model to predict pulmonary embolism in cardiology suspected patients: A 10-year retrospective analysis
  27. Downregulation of lncRNA XLOC_032768 in diabetic patients predicts the occurrence of diabetic nephropathy
  28. Circ_0051428 targeting miR-885-3p/MMP2 axis enhances the malignancy of cervical cancer
  29. Effectiveness of ginkgo diterpene lactone meglumine on cognitive function in patients with acute ischemic stroke
  30. The construction of a novel prognostic prediction model for glioma based on GWAS-identified prognostic-related risk loci
  31. Evaluating the impact of childhood BMI on the risk of coronavirus disease 2019: A Mendelian randomization study
  32. Lactate dehydrogenase to albumin ratio is associated with in-hospital mortality in patients with acute heart failure: Data from the MIMIC-III database
  33. CD36-mediated podocyte lipotoxicity promotes foot process effacement
  34. Efficacy of etonogestrel subcutaneous implants versus the levonorgestrel-releasing intrauterine system in the conservative treatment of adenomyosis
  35. FLRT2 mediates chondrogenesis of nasal septal cartilage and mandibular condyle cartilage
  36. Challenges in treating primary immune thrombocytopenia patients undergoing COVID-19 vaccination: A retrospective study
  37. Let-7 family regulates HaCaT cell proliferation and apoptosis via the ΔNp63/PI3K/AKT pathway
  38. Phospholipid transfer protein ameliorates sepsis-induced cardiac dysfunction through NLRP3 inflammasome inhibition
  39. Postoperative cognitive dysfunction in elderly patients with colorectal cancer: A randomized controlled study comparing goal-directed and conventional fluid therapy
  40. Long-pulsed ultrasound-mediated microbubble thrombolysis in a rat model of microvascular obstruction
  41. High SEC61A1 expression predicts poor outcome of acute myeloid leukemia
  42. Comparison of polymerase chain reaction and next-generation sequencing with conventional urine culture for the diagnosis of urinary tract infections: A meta-analysis
  43. Secreted frizzled-related protein 5 protects against renal fibrosis by inhibiting Wnt/β-catenin pathway
  44. Pan-cancer and single-cell analysis of actin cytoskeleton genes related to disulfidptosis
  45. Overexpression of miR-532-5p restrains oxidative stress response of chondrocytes in nontraumatic osteonecrosis of the femoral head by inhibiting ABL1
  46. Autologous liver transplantation for unresectable hepatobiliary malignancies in enhanced recovery after surgery model
  47. Clinical analysis of incomplete rupture of the uterus secondary to previous cesarean section
  48. Abnormal sleep duration is associated with sarcopenia in older Chinese people: A large retrospective cross-sectional study
  49. No genetic causality between obesity and benign paroxysmal vertigo: A two-sample Mendelian randomization study
  50. Identification and validation of autophagy-related genes in SSc
  51. Long non-coding RNA SRA1 suppresses radiotherapy resistance in esophageal squamous cell carcinoma by modulating glycolytic reprogramming
  52. Evaluation of quality of life in patients with schizophrenia: An inpatient social welfare institution-based cross-sectional study
  53. The possible role of oxidative stress marker glutathione in the assessment of cognitive impairment in multiple sclerosis
  54. Compilation of a self-management assessment scale for postoperative patients with aortic dissection
  55. Left atrial appendage closure in conjunction with radiofrequency ablation: Effects on left atrial functioning in patients with paroxysmal atrial fibrillation
  56. Effect of anterior femoral cortical notch grade on postoperative function and complications during TKA surgery: A multicenter, retrospective study
  57. Clinical characteristics and assessment of risk factors in patients with influenza A-induced severe pneumonia after the prevalence of SARS-CoV-2
  58. Analgesia nociception index is an indicator of laparoscopic trocar insertion-induced transient nociceptive stimuli
  59. High STAT4 expression correlates with poor prognosis in acute myeloid leukemia and facilitates disease progression by upregulating VEGFA expression
  60. Factors influencing cardiovascular system-related post-COVID-19 sequelae: A single-center cohort study
  61. HOXD10 regulates intestinal permeability and inhibits inflammation of dextran sulfate sodium-induced ulcerative colitis through the inactivation of the Rho/ROCK/MMPs axis
  62. Mesenchymal stem cell-derived exosomal miR-26a induces ferroptosis, suppresses hepatic stellate cell activation, and ameliorates liver fibrosis by modulating SLC7A11
  63. Endovascular thrombectomy versus intravenous thrombolysis for primary distal, medium vessel occlusion in acute ischemic stroke
  64. ANO6 (TMEM16F) inhibits gastrointestinal stromal tumor growth and induces ferroptosis
  65. Prognostic value of EIF5A2 in solid tumors: A meta-analysis and bioinformatics analysis
  66. The role of enhanced expression of Cx43 in patients with ulcerative colitis
  67. Choosing a COVID-19 vaccination site might be driven by anxiety and body vigilance
  68. Role of ICAM-1 in triple-negative breast cancer
  69. Cost-effectiveness of ambroxol in the treatment of Gaucher disease type 2
  70. HLA-DRB5 promotes immune thrombocytopenia via activating CD8+ T cells
  71. Efficacy and factors of myofascial release therapy combined with electrical and magnetic stimulation in the treatment of chronic pelvic pain syndrome
  72. Efficacy of tacrolimus monotherapy in primary membranous nephropathy
  73. Mechanisms of Tripterygium wilfordii Hook F on treating rheumatoid arthritis explored by network pharmacology analysis and molecular docking
  74. FBXO45 levels regulated ferroptosis renal tubular epithelial cells in a model of diabetic nephropathy by PLK1
  75. Optimizing anesthesia strategies to NSCLC patients in VATS procedures: Insights from drug requirements and patient recovery patterns
  76. Alpha-lipoic acid upregulates the PPARγ/NRF2/GPX4 signal pathway to inhibit ferroptosis in the pathogenesis of unexplained recurrent pregnancy loss
  77. Correlation between fat-soluble vitamin levels and inflammatory factors in paediatric community-acquired pneumonia: A prospective study
  78. CD1d affects the proliferation, migration, and apoptosis of human papillary thyroid carcinoma TPC-1 cells via regulating MAPK/NF-κB signaling pathway
  79. miR-let-7a inhibits sympathetic nerve remodeling after myocardial infarction by downregulating the expression of nerve growth factor
  80. Immune response analysis of solid organ transplantation recipients inoculated with inactivated COVID-19 vaccine: A retrospective analysis
  81. The H2Valdien derivatives regulate the epithelial–mesenchymal transition of hepatoma carcinoma cells through the Hedgehog signaling pathway
  82. Clinical efficacy of dexamethasone combined with isoniazid in the treatment of tuberculous meningitis and its effect on peripheral blood T cell subsets
  83. Comparison of short-segment and long-segment fixation in treatment of degenerative scoliosis and analysis of factors associated with adjacent spondylolisthesis
  84. Lycopene inhibits pyroptosis of endothelial progenitor cells induced by ox-LDL through the AMPK/mTOR/NLRP3 pathway
  85. Methylation regulation for FUNDC1 stability in childhood leukemia was up-regulated and facilitates metastasis and reduces ferroptosis of leukemia through mitochondrial damage by FBXL2
  86. Correlation of single-fiber electromyography studies and functional status in patients with amyotrophic lateral sclerosis
  87. Risk factors of postoperative airway obstruction complications in children with oral floor mass
  88. Expression levels and clinical significance of serum miR-19a/CCL20 in patients with acute cerebral infarction
  89. Physical activity and mental health trends in Korean adolescents: Analyzing the impact of the COVID-19 pandemic from 2018 to 2022
  90. Evaluating anemia in HIV-infected patients using chest CT
  91. Ponticulus posticus and skeletal malocclusion: A pilot study in a Southern Italian pre-orthodontic court
  92. Causal association of circulating immune cells and lymphoma: A Mendelian randomization study
  93. Assessment of the renal function and fibrosis indexes of conventional western medicine with Chinese medicine for dredging collaterals on treating renal fibrosis: A systematic review and meta-analysis
  94. Comprehensive landscape of integrator complex subunits and their association with prognosis and tumor microenvironment in gastric cancer
  95. New target-HMGCR inhibitors for the treatment of primary sclerosing cholangitis: A drug Mendelian randomization study
  96. Population pharmacokinetics of meropenem in critically ill patients
  97. Comparison of the ability of newly inflammatory markers to predict complicated appendicitis
  98. Comparative morphology of the cruciate ligaments: A radiological study
  99. Immune landscape of hepatocellular carcinoma: The central role of TP53-inducible glycolysis and apoptosis regulator
  100. Serum SIRT3 levels in epilepsy patients and its association with clinical outcomes and severity: A prospective observational study
  101. SHP-1 mediates cigarette smoke extract-induced epithelial–mesenchymal transformation and inflammation in 16HBE cells
  102. Acute hyper-hypoxia accelerates the development of depression in mice via the IL-6/PGC1α/MFN2 signaling pathway
  103. The GJB3 correlates with the prognosis, immune cell infiltration, and therapeutic responses in lung adenocarcinoma
  104. Physical fitness and blood parameters outcomes of breast cancer survivor in a low-intensity circuit resistance exercise program
  105. Exploring anesthetic-induced gene expression changes and immune cell dynamics in atrial tissue post-coronary artery bypass graft surgery
  106. Empagliflozin improves aortic injury in obese mice by regulating fatty acid metabolism
  107. Analysis of the risk factors of the radiation-induced encephalopathy in nasopharyngeal carcinoma: A retrospective cohort study
  108. Reproductive outcomes in women with BRCA 1/2 germline mutations: A retrospective observational study and literature review
  109. Evaluation of upper airway ultrasonographic measurements in predicting difficult intubation: A cross-section of the Turkish population
  110. Prognostic and diagnostic value of circulating IGFBP2 in pancreatic cancer
  111. Postural stability after operative reconstruction of the AFTL in chronic ankle instability comparing three different surgical techniques
  112. Research trends related to emergence agitation in the post-anaesthesia care unit from 2001 to 2023: A bibliometric analysis
  113. Frequency and clinicopathological correlation of gastrointestinal polyps: A six-year single center experience
  114. ACSL4 mediates inflammatory bowel disease and contributes to LPS-induced intestinal epithelial cell dysfunction by activating ferroptosis and inflammation
  115. Affibody-based molecular probe 99mTc-(HE)3ZHER2:V2 for non-invasive HER2 detection in ovarian and breast cancer xenografts
  116. Effectiveness of nutritional support for clinical outcomes in gastric cancer patients: A meta-analysis of randomized controlled trials
  117. The relationship between IFN-γ, IL-10, IL-6 cytokines, and severity of the condition with serum zinc and Fe in children infected with Mycoplasma pneumoniae
  118. Paraquat disrupts the blood–brain barrier by increasing IL-6 expression and oxidative stress through the activation of PI3K/AKT signaling pathway
  119. Sleep quality associate with the increased prevalence of cognitive impairment in coronary artery disease patients: A retrospective case–control study
  120. Dioscin protects against chronic prostatitis through the TLR4/NF-κB pathway
  121. Association of polymorphisms in FBN1, MYH11, and TGF-β signaling-related genes with susceptibility of sporadic thoracic aortic aneurysm and dissection in the Zhejiang Han population
  122. Application value of multi-parameter magnetic resonance image-transrectal ultrasound cognitive fusion in prostate biopsy
  123. Laboratory variables‐based artificial neural network models for predicting fatty liver disease: A retrospective study
  124. Decreased BIRC5-206 promotes epithelial–mesenchymal transition in nasopharyngeal carcinoma through sponging miR-145-5p
  125. Sepsis induces the cardiomyocyte apoptosis and cardiac dysfunction through activation of YAP1/Serpine1/caspase-3 pathway
  126. Assessment of iron metabolism and iron deficiency in incident patients on incident continuous ambulatory peritoneal dialysis
  127. Tibial periosteum flap combined with autologous bone grafting in the treatment of Gustilo-IIIB/IIIC open tibial fractures
  128. The application of intravenous general anesthesia under nasopharyngeal airway assisted ventilation undergoing ureteroscopic holmium laser lithotripsy: A prospective, single-center, controlled trial
  129. Long intergenic noncoding RNA for IGF2BP2 stability suppresses gastric cancer cell apoptosis by inhibiting the maturation of microRNA-34a
  130. Role of FOXM1 and AURKB in regulating keratinocyte function in psoriasis
  131. Parental control attitudes over their pre-school children’s diet
  132. The role of auto-HSCT in extranodal natural killer/T cell lymphoma
  133. Significance of negative cervical cytology and positive HPV in the diagnosis of cervical lesions by colposcopy
  134. Echinacoside inhibits PASMCs calcium overload to prevent hypoxic pulmonary artery remodeling by regulating TRPC1/4/6 and calmodulin
  135. ADAR1 plays a protective role in proximal tubular cells under high glucose conditions by attenuating the PI3K/AKT/mTOR signaling pathway
  136. The risk of cancer among insulin glargine users in Lithuania: A retrospective population-based study
  137. The unusual location of primary hydatid cyst: A case series study
  138. Intraoperative changes in electrophysiological monitoring can be used to predict clinical outcomes in patients with spinal cavernous malformation
  139. Obesity and risk of placenta accreta spectrum: A meta-analysis
  140. Shikonin alleviates asthma phenotypes in mice via an airway epithelial STAT3-dependent mechanism
  141. NSUN6 and HTR7 disturbed the stability of carotid atherosclerotic plaques by regulating the immune responses of macrophages
  142. The effect of COVID-19 lockdown on admission rates in Maternity Hospital
  143. Temporal muscle thickness is not a prognostic predictor in patients with high-grade glioma, an experience at two centers in China
  144. Luteolin alleviates cerebral ischemia/reperfusion injury by regulating cell pyroptosis
  145. Therapeutic role of respiratory exercise in patients with tuberculous pleurisy
  146. Effects of CFTR-ENaC on spinal cord edema after spinal cord injury
  147. Irisin-regulated lncRNAs and their potential regulatory functions in chondrogenic differentiation of human mesenchymal stem cells
  148. DMD mutations in pediatric patients with phenotypes of Duchenne/Becker muscular dystrophy
  149. Combination of C-reactive protein and fibrinogen-to-albumin ratio as a novel predictor of all-cause mortality in heart failure patients
  150. Significant role and the underly mechanism of cullin-1 in chronic obstructive pulmonary disease
  151. Ferroptosis-related prognostic model of mantle cell lymphoma
  152. Observation of choking reaction and other related indexes in elderly painless fiberoptic bronchoscopy with transnasal high-flow humidification oxygen therapy
  153. A bibliometric analysis of Prader-Willi syndrome from 2002 to 2022
  154. The causal effects of childhood sunburn occasions on melanoma: A univariable and multivariable Mendelian randomization study
  155. Oxidative stress regulates glycogen synthase kinase-3 in lymphocytes of diabetes mellitus patients complicated with cerebral infarction
  156. Role of COX6C and NDUFB3 in septic shock and stroke
  157. Trends in disease burden of type 2 diabetes, stroke, and hypertensive heart disease attributable to high BMI in China: 1990–2019
  158. Purinergic P2X7 receptor mediates hyperoxia-induced injury in pulmonary microvascular endothelial cells via NLRP3-mediated pyroptotic pathway
  159. Investigating the role of oviductal mucosa–endometrial co-culture in modulating factors relevant to embryo implantation
  160. Analgesic effect of external oblique intercostal block in laparoscopic cholecystectomy: A retrospective study
  161. Elevated serum miR-142-5p correlates with ischemic lesions and both NSE and S100β in ischemic stroke patients
  162. Correlation between the mechanism of arteriopathy in IgA nephropathy and blood stasis syndrome: A cohort study
  163. Risk factors for progressive kyphosis after percutaneous kyphoplasty in osteoporotic vertebral compression fracture
  164. Predictive role of neuron-specific enolase and S100-β in early neurological deterioration and unfavorable prognosis in patients with ischemic stroke
  165. The potential risk factors of postoperative cognitive dysfunction for endovascular therapy in acute ischemic stroke with general anesthesia
  166. Fluoxetine inhibited RANKL-induced osteoclastic differentiation in vitro
  167. Detection of serum FOXM1 and IGF2 in patients with ARDS and their correlation with disease and prognosis
  168. Rhein promotes skin wound healing by activating the PI3K/AKT signaling pathway
  169. Differences in mortality risk by levels of physical activity among persons with disabilities in South Korea
  170. Review Articles
  171. Cutaneous signs of selected cardiovascular disorders: A narrative review
  172. XRCC1 and hOGG1 polymorphisms and endometrial carcinoma: A meta-analysis
  173. A narrative review on adverse drug reactions of COVID-19 treatments on the kidney
  174. Emerging role and function of SPDL1 in human health and diseases
  175. Adverse reactions of piperacillin: A literature review of case reports
  176. Molecular mechanism and intervention measures of microvascular complications in diabetes
  177. Regulation of mesenchymal stem cell differentiation by autophagy
  178. Molecular landscape of borderline ovarian tumours: A systematic review
  179. Advances in synthetic lethality modalities for glioblastoma multiforme
  180. Investigating hormesis, aging, and neurodegeneration: From bench to clinics
  181. Frankincense: A neuronutrient to approach Parkinson’s disease treatment
  182. Sox9: A potential regulator of cancer stem cells in osteosarcoma
  183. Early detection of cardiovascular risk markers through non-invasive ultrasound methodologies in periodontitis patients
  184. Advanced neuroimaging and criminal interrogation in lie detection
  185. Maternal factors for neural tube defects in offspring: An umbrella review
  186. The chemoprotective hormetic effects of rosmarinic acid
  187. CBD’s potential impact on Parkinson’s disease: An updated overview
  188. Progress in cytokine research for ARDS: A comprehensive review
  189. Utilizing reactive oxygen species-scavenging nanoparticles for targeting oxidative stress in the treatment of ischemic stroke: A review
  190. NRXN1-related disorders, attempt to better define clinical assessment
  191. Lidocaine infusion for the treatment of complex regional pain syndrome: Case series and literature review
  192. Trends and future directions of autophagy in osteosarcoma: A bibliometric analysis
  193. Iron in ventricular remodeling and aneurysms post-myocardial infarction
  194. Case Reports
  195. Sirolimus potentiated angioedema: A case report and review of the literature
  196. Identification of mixed anaerobic infections after inguinal hernia repair based on metagenomic next-generation sequencing: A case report
  197. Successful treatment with bortezomib in combination with dexamethasone in a middle-aged male with idiopathic multicentric Castleman’s disease: A case report
  198. Complete heart block associated with hepatitis A infection in a female child with fatal outcome
  199. Elevation of D-dimer in eosinophilic gastrointestinal diseases in the absence of venous thrombosis: A case series and literature review
  200. Four years of natural progressive course: A rare case report of juvenile Xp11.2 translocations renal cell carcinoma with TFE3 gene fusion
  201. Advancing prenatal diagnosis: Echocardiographic detection of Scimitar syndrome in China – A case series
  202. Outcomes and complications of hemodialysis in patients with renal cancer following bilateral nephrectomy
  203. Anti-HMGCR myopathy mimicking facioscapulohumeral muscular dystrophy
  204. Recurrent opportunistic infections in a HIV-negative patient with combined C6 and NFKB1 mutations: A case report, pedigree analysis, and literature review
  205. Letter to the Editor
  206. Letter to the Editor: Total parenteral nutrition-induced Wernicke’s encephalopathy after oncologic gastrointestinal surgery
  207. Erratum
  208. Erratum to “Bladder-embedded ectopic intrauterine device with calculus”
  209. Retraction
  210. Retraction of “XRCC1 and hOGG1 polymorphisms and endometrial carcinoma: A meta-analysis”
  211. Corrigendum
  212. Corrigendum to “Investigating hormesis, aging, and neurodegeneration: From bench to clinics”
  213. Corrigendum to “Frankincense: A neuronutrient to approach Parkinson’s disease treatment”
  214. Special Issue The evolving saga of RNAs from bench to bedside - Part II
  215. Machine-learning-based prediction of a diagnostic model using autophagy-related genes based on RNA sequencing for patients with papillary thyroid carcinoma
  216. Unlocking the future of hepatocellular carcinoma treatment: A comprehensive analysis of disulfidptosis-related lncRNAs for prognosis and drug screening
  217. Elevated mRNA level indicates FSIP1 promotes EMT and gastric cancer progression by regulating fibroblasts in tumor microenvironment
  218. Special Issue Advancements in oncology: bridging clinical and experimental research - Part I
  219. Ultrasound-guided transperineal vs transrectal prostate biopsy: A meta-analysis of diagnostic accuracy and complication rates
  220. Assessment of diagnostic value of unilateral systematic biopsy combined with targeted biopsy in detecting clinically significant prostate cancer
  221. SENP7 inhibits glioblastoma metastasis and invasion by dissociating SUMO2/3 binding to specific target proteins
  222. MARK1 suppress malignant progression of hepatocellular carcinoma and improves sorafenib resistance through negatively regulating POTEE
  223. Analysis of postoperative complications in bladder cancer patients
  224. Carboplatin combined with arsenic trioxide versus carboplatin combined with docetaxel treatment for LACC: A randomized, open-label, phase II clinical study
  225. Special Issue Exploring the biological mechanism of human diseases based on MultiOmics Technology - Part I
  226. Comprehensive pan-cancer investigation of carnosine dipeptidase 1 and its prospective prognostic significance in hepatocellular carcinoma
  227. Identification of signatures associated with microsatellite instability and immune characteristics to predict the prognostic risk of colon cancer
  228. Single-cell analysis identified key macrophage subpopulations associated with atherosclerosis
https://doi.org/10.1515/med-2024-1031
Keywords for this article
fatty liver disease; artificial neural network; model; prediction; laboratory variables
Creative Commons
BY 4.0

Articles in the same Issue

  1. Research Articles
  2. EDNRB inhibits the growth and migration of prostate cancer cells by activating the cGMP-PKG pathway
  3. STK11 (LKB1) mutation suppresses ferroptosis in lung adenocarcinoma by facilitating monounsaturated fatty acid synthesis
  4. Association of SOX6 gene polymorphisms with Kashin-Beck disease risk in the Chinese Han population
  5. The pyroptosis-related signature predicts prognosis and influences the tumor immune microenvironment in dedifferentiated liposarcoma
  6. METTL3 attenuates ferroptosis sensitivity in lung cancer via modulating TFRC
  7. Identification and validation of molecular subtypes and prognostic signature for stage I and stage II gastric cancer based on neutrophil extracellular traps
  8. Novel lumbar plexus block versus femoral nerve block for analgesia and motor recovery after total knee arthroplasty
  9. Correlation between ABCB1 and OLIG2 polymorphisms and the severity and prognosis of patients with cerebral infarction
  10. Study on the radiotherapy effect and serum neutral granulocyte lymphocyte ratio and inflammatory factor expression of nasopharyngeal carcinoma
  11. Transcriptome analysis of effects of Tecrl deficiency on cardiometabolic and calcium regulation in cardiac tissue
  12. Aflatoxin B1 induces infertility, fetal deformities, and potential therapies
  13. Serum levels of HMW adiponectin and its receptors are associated with cytokine levels and clinical characteristics in chronic obstructive pulmonary disease
  14. METTL3-mediated methylation of CYP2C19 mRNA may aggravate clopidogrel resistance in ischemic stroke patients
  15. Understand how machine learning impact lung cancer research from 2010 to 2021: A bibliometric analysis
  16. Pressure ulcers in German hospitals: Analysis of reimbursement and length of stay
  17. Metformin plus L-carnitine enhances brown/beige adipose tissue activity via Nrf2/HO-1 signaling to reduce lipid accumulation and inflammation in murine obesity
  18. Downregulation of carbonic anhydrase IX expression in mouse xenograft nasopharyngeal carcinoma model via doxorubicin nanobubble combined with ultrasound
  19. Feasibility of 3-dimensional printed models in simulated training and teaching of transcatheter aortic valve replacement
  20. miR-335-3p improves type II diabetes mellitus by IGF-1 regulating macrophage polarization
  21. The analyses of human MCPH1 DNA repair machinery and genetic variations
  22. Activation of Piezo1 increases the sensitivity of breast cancer to hyperthermia therapy
  23. Comprehensive analysis based on the disulfidptosis-related genes identifies hub genes and immune infiltration for pancreatic adenocarcinoma
  24. Changes of serum CA125 and PGE2 before and after high-intensity focused ultrasound combined with GnRH-a in treatment of patients with adenomyosis
  25. The clinical value of the hepatic venous pressure gradient in patients undergoing hepatic resection for hepatocellular carcinoma with or without liver cirrhosis
  26. Development and validation of a novel model to predict pulmonary embolism in cardiology suspected patients: A 10-year retrospective analysis
  27. Downregulation of lncRNA XLOC_032768 in diabetic patients predicts the occurrence of diabetic nephropathy
  28. Circ_0051428 targeting miR-885-3p/MMP2 axis enhances the malignancy of cervical cancer
  29. Effectiveness of ginkgo diterpene lactone meglumine on cognitive function in patients with acute ischemic stroke
  30. The construction of a novel prognostic prediction model for glioma based on GWAS-identified prognostic-related risk loci
  31. Evaluating the impact of childhood BMI on the risk of coronavirus disease 2019: A Mendelian randomization study
  32. Lactate dehydrogenase to albumin ratio is associated with in-hospital mortality in patients with acute heart failure: Data from the MIMIC-III database
  33. CD36-mediated podocyte lipotoxicity promotes foot process effacement
  34. Efficacy of etonogestrel subcutaneous implants versus the levonorgestrel-releasing intrauterine system in the conservative treatment of adenomyosis
  35. FLRT2 mediates chondrogenesis of nasal septal cartilage and mandibular condyle cartilage
  36. Challenges in treating primary immune thrombocytopenia patients undergoing COVID-19 vaccination: A retrospective study
  37. Let-7 family regulates HaCaT cell proliferation and apoptosis via the ΔNp63/PI3K/AKT pathway
  38. Phospholipid transfer protein ameliorates sepsis-induced cardiac dysfunction through NLRP3 inflammasome inhibition
  39. Postoperative cognitive dysfunction in elderly patients with colorectal cancer: A randomized controlled study comparing goal-directed and conventional fluid therapy
  40. Long-pulsed ultrasound-mediated microbubble thrombolysis in a rat model of microvascular obstruction
  41. High SEC61A1 expression predicts poor outcome of acute myeloid leukemia
  42. Comparison of polymerase chain reaction and next-generation sequencing with conventional urine culture for the diagnosis of urinary tract infections: A meta-analysis
  43. Secreted frizzled-related protein 5 protects against renal fibrosis by inhibiting Wnt/β-catenin pathway
  44. Pan-cancer and single-cell analysis of actin cytoskeleton genes related to disulfidptosis
  45. Overexpression of miR-532-5p restrains oxidative stress response of chondrocytes in nontraumatic osteonecrosis of the femoral head by inhibiting ABL1
  46. Autologous liver transplantation for unresectable hepatobiliary malignancies in enhanced recovery after surgery model
  47. Clinical analysis of incomplete rupture of the uterus secondary to previous cesarean section
  48. Abnormal sleep duration is associated with sarcopenia in older Chinese people: A large retrospective cross-sectional study
  49. No genetic causality between obesity and benign paroxysmal vertigo: A two-sample Mendelian randomization study
  50. Identification and validation of autophagy-related genes in SSc
  51. Long non-coding RNA SRA1 suppresses radiotherapy resistance in esophageal squamous cell carcinoma by modulating glycolytic reprogramming
  52. Evaluation of quality of life in patients with schizophrenia: An inpatient social welfare institution-based cross-sectional study
  53. The possible role of oxidative stress marker glutathione in the assessment of cognitive impairment in multiple sclerosis
  54. Compilation of a self-management assessment scale for postoperative patients with aortic dissection
  55. Left atrial appendage closure in conjunction with radiofrequency ablation: Effects on left atrial functioning in patients with paroxysmal atrial fibrillation
  56. Effect of anterior femoral cortical notch grade on postoperative function and complications during TKA surgery: A multicenter, retrospective study
  57. Clinical characteristics and assessment of risk factors in patients with influenza A-induced severe pneumonia after the prevalence of SARS-CoV-2
  58. Analgesia nociception index is an indicator of laparoscopic trocar insertion-induced transient nociceptive stimuli
  59. High STAT4 expression correlates with poor prognosis in acute myeloid leukemia and facilitates disease progression by upregulating VEGFA expression
  60. Factors influencing cardiovascular system-related post-COVID-19 sequelae: A single-center cohort study
  61. HOXD10 regulates intestinal permeability and inhibits inflammation of dextran sulfate sodium-induced ulcerative colitis through the inactivation of the Rho/ROCK/MMPs axis
  62. Mesenchymal stem cell-derived exosomal miR-26a induces ferroptosis, suppresses hepatic stellate cell activation, and ameliorates liver fibrosis by modulating SLC7A11
  63. Endovascular thrombectomy versus intravenous thrombolysis for primary distal, medium vessel occlusion in acute ischemic stroke
  64. ANO6 (TMEM16F) inhibits gastrointestinal stromal tumor growth and induces ferroptosis
  65. Prognostic value of EIF5A2 in solid tumors: A meta-analysis and bioinformatics analysis
  66. The role of enhanced expression of Cx43 in patients with ulcerative colitis
  67. Choosing a COVID-19 vaccination site might be driven by anxiety and body vigilance
  68. Role of ICAM-1 in triple-negative breast cancer
  69. Cost-effectiveness of ambroxol in the treatment of Gaucher disease type 2
  70. HLA-DRB5 promotes immune thrombocytopenia via activating CD8+ T cells
  71. Efficacy and factors of myofascial release therapy combined with electrical and magnetic stimulation in the treatment of chronic pelvic pain syndrome
  72. Efficacy of tacrolimus monotherapy in primary membranous nephropathy
  73. Mechanisms of Tripterygium wilfordii Hook F on treating rheumatoid arthritis explored by network pharmacology analysis and molecular docking
  74. FBXO45 levels regulated ferroptosis renal tubular epithelial cells in a model of diabetic nephropathy by PLK1
  75. Optimizing anesthesia strategies to NSCLC patients in VATS procedures: Insights from drug requirements and patient recovery patterns
  76. Alpha-lipoic acid upregulates the PPARγ/NRF2/GPX4 signal pathway to inhibit ferroptosis in the pathogenesis of unexplained recurrent pregnancy loss
  77. Correlation between fat-soluble vitamin levels and inflammatory factors in paediatric community-acquired pneumonia: A prospective study
  78. CD1d affects the proliferation, migration, and apoptosis of human papillary thyroid carcinoma TPC-1 cells via regulating MAPK/NF-κB signaling pathway
  79. miR-let-7a inhibits sympathetic nerve remodeling after myocardial infarction by downregulating the expression of nerve growth factor
  80. Immune response analysis of solid organ transplantation recipients inoculated with inactivated COVID-19 vaccine: A retrospective analysis
  81. The H2Valdien derivatives regulate the epithelial–mesenchymal transition of hepatoma carcinoma cells through the Hedgehog signaling pathway
  82. Clinical efficacy of dexamethasone combined with isoniazid in the treatment of tuberculous meningitis and its effect on peripheral blood T cell subsets
  83. Comparison of short-segment and long-segment fixation in treatment of degenerative scoliosis and analysis of factors associated with adjacent spondylolisthesis
  84. Lycopene inhibits pyroptosis of endothelial progenitor cells induced by ox-LDL through the AMPK/mTOR/NLRP3 pathway
  85. Methylation regulation for FUNDC1 stability in childhood leukemia was up-regulated and facilitates metastasis and reduces ferroptosis of leukemia through mitochondrial damage by FBXL2
  86. Correlation of single-fiber electromyography studies and functional status in patients with amyotrophic lateral sclerosis
  87. Risk factors of postoperative airway obstruction complications in children with oral floor mass
  88. Expression levels and clinical significance of serum miR-19a/CCL20 in patients with acute cerebral infarction
  89. Physical activity and mental health trends in Korean adolescents: Analyzing the impact of the COVID-19 pandemic from 2018 to 2022
  90. Evaluating anemia in HIV-infected patients using chest CT
  91. Ponticulus posticus and skeletal malocclusion: A pilot study in a Southern Italian pre-orthodontic court
  92. Causal association of circulating immune cells and lymphoma: A Mendelian randomization study
  93. Assessment of the renal function and fibrosis indexes of conventional western medicine with Chinese medicine for dredging collaterals on treating renal fibrosis: A systematic review and meta-analysis
  94. Comprehensive landscape of integrator complex subunits and their association with prognosis and tumor microenvironment in gastric cancer
  95. New target-HMGCR inhibitors for the treatment of primary sclerosing cholangitis: A drug Mendelian randomization study
  96. Population pharmacokinetics of meropenem in critically ill patients
  97. Comparison of the ability of newly inflammatory markers to predict complicated appendicitis
  98. Comparative morphology of the cruciate ligaments: A radiological study
  99. Immune landscape of hepatocellular carcinoma: The central role of TP53-inducible glycolysis and apoptosis regulator
  100. Serum SIRT3 levels in epilepsy patients and its association with clinical outcomes and severity: A prospective observational study
  101. SHP-1 mediates cigarette smoke extract-induced epithelial–mesenchymal transformation and inflammation in 16HBE cells
  102. Acute hyper-hypoxia accelerates the development of depression in mice via the IL-6/PGC1α/MFN2 signaling pathway
  103. The GJB3 correlates with the prognosis, immune cell infiltration, and therapeutic responses in lung adenocarcinoma
  104. Physical fitness and blood parameters outcomes of breast cancer survivor in a low-intensity circuit resistance exercise program
  105. Exploring anesthetic-induced gene expression changes and immune cell dynamics in atrial tissue post-coronary artery bypass graft surgery
  106. Empagliflozin improves aortic injury in obese mice by regulating fatty acid metabolism
  107. Analysis of the risk factors of the radiation-induced encephalopathy in nasopharyngeal carcinoma: A retrospective cohort study
  108. Reproductive outcomes in women with BRCA 1/2 germline mutations: A retrospective observational study and literature review
  109. Evaluation of upper airway ultrasonographic measurements in predicting difficult intubation: A cross-section of the Turkish population
  110. Prognostic and diagnostic value of circulating IGFBP2 in pancreatic cancer
  111. Postural stability after operative reconstruction of the AFTL in chronic ankle instability comparing three different surgical techniques
  112. Research trends related to emergence agitation in the post-anaesthesia care unit from 2001 to 2023: A bibliometric analysis
  113. Frequency and clinicopathological correlation of gastrointestinal polyps: A six-year single center experience
  114. ACSL4 mediates inflammatory bowel disease and contributes to LPS-induced intestinal epithelial cell dysfunction by activating ferroptosis and inflammation
  115. Affibody-based molecular probe 99mTc-(HE)3ZHER2:V2 for non-invasive HER2 detection in ovarian and breast cancer xenografts
  116. Effectiveness of nutritional support for clinical outcomes in gastric cancer patients: A meta-analysis of randomized controlled trials
  117. The relationship between IFN-γ, IL-10, IL-6 cytokines, and severity of the condition with serum zinc and Fe in children infected with Mycoplasma pneumoniae
  118. Paraquat disrupts the blood–brain barrier by increasing IL-6 expression and oxidative stress through the activation of PI3K/AKT signaling pathway
  119. Sleep quality associate with the increased prevalence of cognitive impairment in coronary artery disease patients: A retrospective case–control study
  120. Dioscin protects against chronic prostatitis through the TLR4/NF-κB pathway
  121. Association of polymorphisms in FBN1, MYH11, and TGF-β signaling-related genes with susceptibility of sporadic thoracic aortic aneurysm and dissection in the Zhejiang Han population
  122. Application value of multi-parameter magnetic resonance image-transrectal ultrasound cognitive fusion in prostate biopsy
  123. Laboratory variables‐based artificial neural network models for predicting fatty liver disease: A retrospective study
  124. Decreased BIRC5-206 promotes epithelial–mesenchymal transition in nasopharyngeal carcinoma through sponging miR-145-5p
  125. Sepsis induces the cardiomyocyte apoptosis and cardiac dysfunction through activation of YAP1/Serpine1/caspase-3 pathway
  126. Assessment of iron metabolism and iron deficiency in incident patients on incident continuous ambulatory peritoneal dialysis
  127. Tibial periosteum flap combined with autologous bone grafting in the treatment of Gustilo-IIIB/IIIC open tibial fractures
  128. The application of intravenous general anesthesia under nasopharyngeal airway assisted ventilation undergoing ureteroscopic holmium laser lithotripsy: A prospective, single-center, controlled trial
  129. Long intergenic noncoding RNA for IGF2BP2 stability suppresses gastric cancer cell apoptosis by inhibiting the maturation of microRNA-34a
  130. Role of FOXM1 and AURKB in regulating keratinocyte function in psoriasis
  131. Parental control attitudes over their pre-school children’s diet
  132. The role of auto-HSCT in extranodal natural killer/T cell lymphoma
  133. Significance of negative cervical cytology and positive HPV in the diagnosis of cervical lesions by colposcopy
  134. Echinacoside inhibits PASMCs calcium overload to prevent hypoxic pulmonary artery remodeling by regulating TRPC1/4/6 and calmodulin
  135. ADAR1 plays a protective role in proximal tubular cells under high glucose conditions by attenuating the PI3K/AKT/mTOR signaling pathway
  136. The risk of cancer among insulin glargine users in Lithuania: A retrospective population-based study
  137. The unusual location of primary hydatid cyst: A case series study
  138. Intraoperative changes in electrophysiological monitoring can be used to predict clinical outcomes in patients with spinal cavernous malformation
  139. Obesity and risk of placenta accreta spectrum: A meta-analysis
  140. Shikonin alleviates asthma phenotypes in mice via an airway epithelial STAT3-dependent mechanism
  141. NSUN6 and HTR7 disturbed the stability of carotid atherosclerotic plaques by regulating the immune responses of macrophages
  142. The effect of COVID-19 lockdown on admission rates in Maternity Hospital
  143. Temporal muscle thickness is not a prognostic predictor in patients with high-grade glioma, an experience at two centers in China
  144. Luteolin alleviates cerebral ischemia/reperfusion injury by regulating cell pyroptosis
  145. Therapeutic role of respiratory exercise in patients with tuberculous pleurisy
  146. Effects of CFTR-ENaC on spinal cord edema after spinal cord injury
  147. Irisin-regulated lncRNAs and their potential regulatory functions in chondrogenic differentiation of human mesenchymal stem cells
  148. DMD mutations in pediatric patients with phenotypes of Duchenne/Becker muscular dystrophy
  149. Combination of C-reactive protein and fibrinogen-to-albumin ratio as a novel predictor of all-cause mortality in heart failure patients
  150. Significant role and the underly mechanism of cullin-1 in chronic obstructive pulmonary disease
  151. Ferroptosis-related prognostic model of mantle cell lymphoma
  152. Observation of choking reaction and other related indexes in elderly painless fiberoptic bronchoscopy with transnasal high-flow humidification oxygen therapy
  153. A bibliometric analysis of Prader-Willi syndrome from 2002 to 2022
  154. The causal effects of childhood sunburn occasions on melanoma: A univariable and multivariable Mendelian randomization study
  155. Oxidative stress regulates glycogen synthase kinase-3 in lymphocytes of diabetes mellitus patients complicated with cerebral infarction
  156. Role of COX6C and NDUFB3 in septic shock and stroke
  157. Trends in disease burden of type 2 diabetes, stroke, and hypertensive heart disease attributable to high BMI in China: 1990–2019
  158. Purinergic P2X7 receptor mediates hyperoxia-induced injury in pulmonary microvascular endothelial cells via NLRP3-mediated pyroptotic pathway
  159. Investigating the role of oviductal mucosa–endometrial co-culture in modulating factors relevant to embryo implantation
  160. Analgesic effect of external oblique intercostal block in laparoscopic cholecystectomy: A retrospective study
  161. Elevated serum miR-142-5p correlates with ischemic lesions and both NSE and S100β in ischemic stroke patients
  162. Correlation between the mechanism of arteriopathy in IgA nephropathy and blood stasis syndrome: A cohort study
  163. Risk factors for progressive kyphosis after percutaneous kyphoplasty in osteoporotic vertebral compression fracture
  164. Predictive role of neuron-specific enolase and S100-β in early neurological deterioration and unfavorable prognosis in patients with ischemic stroke
  165. The potential risk factors of postoperative cognitive dysfunction for endovascular therapy in acute ischemic stroke with general anesthesia
  166. Fluoxetine inhibited RANKL-induced osteoclastic differentiation in vitro
  167. Detection of serum FOXM1 and IGF2 in patients with ARDS and their correlation with disease and prognosis
  168. Rhein promotes skin wound healing by activating the PI3K/AKT signaling pathway
  169. Differences in mortality risk by levels of physical activity among persons with disabilities in South Korea
  170. Review Articles
  171. Cutaneous signs of selected cardiovascular disorders: A narrative review
  172. XRCC1 and hOGG1 polymorphisms and endometrial carcinoma: A meta-analysis
  173. A narrative review on adverse drug reactions of COVID-19 treatments on the kidney
  174. Emerging role and function of SPDL1 in human health and diseases
  175. Adverse reactions of piperacillin: A literature review of case reports
  176. Molecular mechanism and intervention measures of microvascular complications in diabetes
  177. Regulation of mesenchymal stem cell differentiation by autophagy
  178. Molecular landscape of borderline ovarian tumours: A systematic review
  179. Advances in synthetic lethality modalities for glioblastoma multiforme
  180. Investigating hormesis, aging, and neurodegeneration: From bench to clinics
  181. Frankincense: A neuronutrient to approach Parkinson’s disease treatment
  182. Sox9: A potential regulator of cancer stem cells in osteosarcoma
  183. Early detection of cardiovascular risk markers through non-invasive ultrasound methodologies in periodontitis patients
  184. Advanced neuroimaging and criminal interrogation in lie detection
  185. Maternal factors for neural tube defects in offspring: An umbrella review
  186. The chemoprotective hormetic effects of rosmarinic acid
  187. CBD’s potential impact on Parkinson’s disease: An updated overview
  188. Progress in cytokine research for ARDS: A comprehensive review
  189. Utilizing reactive oxygen species-scavenging nanoparticles for targeting oxidative stress in the treatment of ischemic stroke: A review
  190. NRXN1-related disorders, attempt to better define clinical assessment
  191. Lidocaine infusion for the treatment of complex regional pain syndrome: Case series and literature review
  192. Trends and future directions of autophagy in osteosarcoma: A bibliometric analysis
  193. Iron in ventricular remodeling and aneurysms post-myocardial infarction
  194. Case Reports
  195. Sirolimus potentiated angioedema: A case report and review of the literature
  196. Identification of mixed anaerobic infections after inguinal hernia repair based on metagenomic next-generation sequencing: A case report
  197. Successful treatment with bortezomib in combination with dexamethasone in a middle-aged male with idiopathic multicentric Castleman’s disease: A case report
  198. Complete heart block associated with hepatitis A infection in a female child with fatal outcome
  199. Elevation of D-dimer in eosinophilic gastrointestinal diseases in the absence of venous thrombosis: A case series and literature review
  200. Four years of natural progressive course: A rare case report of juvenile Xp11.2 translocations renal cell carcinoma with TFE3 gene fusion
  201. Advancing prenatal diagnosis: Echocardiographic detection of Scimitar syndrome in China – A case series
  202. Outcomes and complications of hemodialysis in patients with renal cancer following bilateral nephrectomy
  203. Anti-HMGCR myopathy mimicking facioscapulohumeral muscular dystrophy
  204. Recurrent opportunistic infections in a HIV-negative patient with combined C6 and NFKB1 mutations: A case report, pedigree analysis, and literature review
  205. Letter to the Editor
  206. Letter to the Editor: Total parenteral nutrition-induced Wernicke’s encephalopathy after oncologic gastrointestinal surgery
  207. Erratum
  208. Erratum to “Bladder-embedded ectopic intrauterine device with calculus”
  209. Retraction
  210. Retraction of “XRCC1 and hOGG1 polymorphisms and endometrial carcinoma: A meta-analysis”
  211. Corrigendum
  212. Corrigendum to “Investigating hormesis, aging, and neurodegeneration: From bench to clinics”
  213. Corrigendum to “Frankincense: A neuronutrient to approach Parkinson’s disease treatment”
  214. Special Issue The evolving saga of RNAs from bench to bedside - Part II
  215. Machine-learning-based prediction of a diagnostic model using autophagy-related genes based on RNA sequencing for patients with papillary thyroid carcinoma
  216. Unlocking the future of hepatocellular carcinoma treatment: A comprehensive analysis of disulfidptosis-related lncRNAs for prognosis and drug screening
  217. Elevated mRNA level indicates FSIP1 promotes EMT and gastric cancer progression by regulating fibroblasts in tumor microenvironment
  218. Special Issue Advancements in oncology: bridging clinical and experimental research - Part I
  219. Ultrasound-guided transperineal vs transrectal prostate biopsy: A meta-analysis of diagnostic accuracy and complication rates
  220. Assessment of diagnostic value of unilateral systematic biopsy combined with targeted biopsy in detecting clinically significant prostate cancer
  221. SENP7 inhibits glioblastoma metastasis and invasion by dissociating SUMO2/3 binding to specific target proteins
  222. MARK1 suppress malignant progression of hepatocellular carcinoma and improves sorafenib resistance through negatively regulating POTEE
  223. Analysis of postoperative complications in bladder cancer patients
  224. Carboplatin combined with arsenic trioxide versus carboplatin combined with docetaxel treatment for LACC: A randomized, open-label, phase II clinical study
  225. Special Issue Exploring the biological mechanism of human diseases based on MultiOmics Technology - Part I
  226. Comprehensive pan-cancer investigation of carnosine dipeptidase 1 and its prospective prognostic significance in hepatocellular carcinoma
  227. Identification of signatures associated with microsatellite instability and immune characteristics to predict the prognostic risk of colon cancer
  228. Single-cell analysis identified key macrophage subpopulations associated with atherosclerosis
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 20.9.2025 from https://www.degruyterbrill.com/document/doi/10.1515/med-2024-1031/html?lang=en
Scroll to top button