Objective To review the progress of artificial intelligence (AI) and radiomics in the study of abdominal aortic aneurysm (AAA). Method The literatures related to AI, radiomics and AAA research in recent years were collected and summarized in detail. Results AI and radiomics influenced AAA research and clinical decisions in terms of feature extraction, risk prediction, patient management, simulation of stent-graft deployment, and data mining. Conclusion The application of AI and radiomics provides new ideas for AAA research and clinical decisions, and is expected to suggest personalized treatment and follow-up protocols to guide clinical practice, aiming to achieve precision medicine of AAA.
ObjectiveTo analyze the correlation between the molecular biological information of SMARCA4-deficient non-small cell lung cancer (SMARCA4-dNSCLC) and its clinical prognosis, and to explore the spatial features and molecular mechanisms of interactions between cells in the tumor microenvironment (TME) of SMARCA4-dNSCLC. MethodsUsing data from The Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO), this study conducted functional enrichment analysis on differentially expressed genes (DEGs) in SMARCA4-dNSCLC and depicted its genomic variation landscape. Through weighted gene co-expression network analysis (WGCNA) and a combination of 10 different machine learning algorithms, patients in the training group were divided into a low-risk group and a high-risk group based on a median risk score (RiskScore). A corresponding prognostic prediction model was established, and on this basis, a nomogram was constructed to predict the 1, 3, and 5-year survival rates of patients. K-M survival curves, receiver operating characteristic (ROC) curves, and time-dependent ROC curves were drawn to evaluate the predictive ability of the model. External datasets from GEO further validated the prognostic value of the prediction model. In addition, we also evaluated the immunological characteristics of the TME of the prognostic model. Finally, using single-cell RNA sequencing (scRNA-seq) and spatial transcriptome (ST), we explored the spatial features of interactions between cells in the TME of SMARCA4-dNSCLC, intercellular communication, and molecular mechanisms. ResultsA total of 56 patients were included in the training group, including 38 males and 18 females, with a median age of 62 (56-70) years. There were 28 patients in both the low-risk and high-risk groups. A total of 474 patients were included in the training group, including 265 males and 209 females, with a median age of 65 (58-70) years. A risk score model composed of 8 prognostic feature genes (ELANE, FSIP2, GFI1B, GPR37, KRT81, RHOV, RP1, SPIC) was established. Compared with patients in the low-risk group, those in the high-risk group showed a more unfavorable prognostic outcome. Immunological feature analysis revealed differences in the infiltration of various immune cells between the low-risk and high-risk groups. ScRNA-seq and ST analyses found that interactions between cells were mainly through macrophage migration inhibitory factor (MIF) signaling pathways (MIF-CD74+CXCR4 and MIF-CD74+CD44) via ligand-receptor pairs, while also describing the niche interactions of the MIF signaling pathway in tissue regions. ConclusionThe 8-gene prognostic model constructed in this study has certain predictive accuracy in predicting the survival of SMARCA4-dNSCLC. Combining the ScRNA-seq and ST analyses, cell-to-cell crosstalk and spatial niche interaction may occur between cells in the TME via the MIF signaling pathway (MIF-CD74+CXCR4 and MIF-CD74+CD44).
With the increasing availability of clinical and biomedical big data, machine learning is being widely used in scientific research and academic papers. It integrates various types of information to predict individual health outcomes. However, deficiencies in reporting key information have gradually emerged. These include issues like data bias, model fairness across different groups, and problems with data quality and applicability. Maintaining predictive accuracy and interpretability in real-world clinical settings is also a challenge. This increases the complexity of safely and effectively applying predictive models to clinical practice. To address these problems, TRIPOD+AI (transparent reporting of a multivariable prediction model for individual prognosis or diagnosis+artificial intelligence) introduces a reporting standard for machine learning models. It is based on TRIPOD and aims to improve transparency, reproducibility, and health equity. These improvements enhance the quality of machine learning model applications. Currently, research on prediction models based on machine learning is rapidly increasing. To help domestic readers better understand and apply TRIPOD+AI, we provide examples and interpretations. We hope this will support researchers in improving the quality of their reports.
Acute respiratory distress syndrome (ARDS) is a serious threat to human life and health disease, with acute onset and high mortality. The current diagnosis of the disease depends on blood gas analysis results, while calculating the oxygenation index. However, blood gas analysis is an invasive operation, and can’t continuously monitor the development of the disease. In response to the above problems, in this study, we proposed a new algorithm for identifying the severity of ARDS disease. Based on a variety of non-invasive physiological parameters of patients, combined with feature selection techniques, this paper sorts the importance of various physiological parameters. The cross-validation technique was used to evaluate the identification performance. The classification results of four supervised learning algorithms using neural network, logistic regression, AdaBoost and Bagging were compared under different feature subsets. The optimal feature subset and classification algorithm are comprehensively selected by the sensitivity, specificity, accuracy and area under curve (AUC) of different algorithms under different feature subsets. We use four supervised learning algorithms to distinguish the severity of ARDS (P/F ≤ 300). The performance of the algorithm is evaluated according to AUC. When AdaBoost uses 20 features, AUC = 0.832 1, the accuracy is 74.82%, and the optimal AUC is obtained. The performance of the algorithm is evaluated according to the number of features. When using 2 features, Bagging has AUC = 0.819 4 and the accuracy is 73.01%. Compared with traditional methods, this method has the advantage of continuously monitoring the development of patients with ARDS and providing medical staff with auxiliary diagnosis suggestions.
Cardiovascular diseases are the leading cause of death and their diagnosis and treatment rely heavily on the variety of clinical data. With the advent of the era of medical big data, artificial intelligence (AI) has been widely applied in many aspects such as imaging, diagnosis and prognosis prediction in cardiovascular medicine, providing a new method for accurate diagnosis and treatment. This paper reviews the application of AI in cardiovascular medicine.
The development and validation of clinical prediction models based on artificial intelligence (AI) and machine learning (ML) methods have become increasingly widespread. However, the prediction model bias risk and applicability evaluation tool developed in 2019 (i.e., PROBAST-2019) has shown significant limitations. Therefore, an expanded and updated version of the PROBAST-2019 tool was released in 2025, known as the PROBAST+AI tool. The tool is divided into two parts including model development and model evaluation. It aims to comprehensively and systematically evaluate potential methodological quality issues in model development, bias risks in model evaluation, and the applicability of models, regardless of the modeling method used. This paper provides a systematic interpretation of the PROBAST+AI tool's items and case analyses, with the aim of guiding and assisting researchers engaged in related studies and promoting the high-quality development of clinical predictive model research.
Objective To investigate the network reorganization and dynamic brain activity in visuospatial neglect (VSN) patients using resting-state electroencephalography (rEEG), and to develop classification models to facilitate its identification. Methods In this retrospective study, stroke patients admitted to the Department of Rehabilitation, Xuanwu Hospital, Capital Medical University between August 2022 and December 2024 were included and divided into VSN (n=22) and non-VSN (n=21) groups based on paper-and-pencil assessments. A healthy control group (n=20) was also recruited. Microstate segmentation and graph-theoretical analysis were applied to rEEG data to extract microstate parameters and topological network features. Four machine learning models (logistic regression, naïve Bayes, k-nearest neighbors, and decision tree) were built for classification. Results Compared with the non-VSN group, the VSN group showed significantly increased mean duration and time coverage in microstate C, and significantly decreased coverage and occurrence in microstate D. Graph-theoretical analysis revealed higher average clustering coefficients in the VSN group. Degree centrality in the frontal-central regions (C1, CZ) was significantly lower, while that in the parietal-occipital regions (P5, P3, PO7, PO5) was significantly higher than in the non-VSN group. Among the classification models, logistic regression and naïve Bayes models performed best, with the mean duration of microstate C contributing most to classification performance. Conclusions Patients with VSN exhibit distinct alterations in electroencephalography microstate dynamics and functional network topology. Microstate parameters play a crucial role in distinguishing VSN from non-VSN stroke cases, and combining these features with machine learning offers a promising approach for early identification and personalized intervention of VSN.
Motor imagery (MI), motion intention of the specific body without actual movements, has attracted wide attention in fields as neuroscience. Classification algorithms for motor imagery electroencephalogram (MI-EEG) signals are able to distinguish different MI tasks based on the physiological information contained by the EEG signals, especially the features extracted from them. In recent years, there have been some new advances in classification algorithms for MI-EEG signals in terms of classifiers versus machine learning strategies. In terms of classifiers, traditional machine learning classifiers have been improved by some researchers, deep learning and Riemannian geometry classifiers have been widely applied as well. In terms of machine learning strategies, ensemble learning, adaptive learning, and transfer learning strategies have been utilized to improve classification accuracies or reach other targets. This paper reviewed the progress of classification algorithms for MI-EEG signals, summarized and evaluated the existing classifiers and machine learning strategies, to provide new ideas for developing classification algorithms with higher performance.
Objective To develop and compare the predictive performance of five machine learning models for adverse postoperative outcomes in cardiac surgery patients, and to identify key decision factors through SHapley Additive exPlanations (SHAP) interpretability analysis. Methods A retrospective collection of perioperative data (including demographic information, preoperative, intraoperative, and postoperative indicators) with 88 variables was conducted from adult cardiac surgery patients at the First Affiliated Hospital of Xinjiang Medical University in 2023. Adverse postoperative outcomes were defined as the occurrence of acute kidney injury and/or in-hospital mortality during the postoperative hospitalization period following cardiac surgery. Patients were divided into an adverse outcome group and a favorable outcome group based on the presence of adverse postoperative outcomes. After screening feature variables using the least absolute shrinkage and selection operator (LASSO) regression method, five machine learning models were constructed: eXtreme gradient boosting (XGBoost), random forest (RF), gradient boosting machine (GBM), light gradient boosting machine (LightGBM), and generalized linear model (GLM). The dataset was randomly divided into a training set and a test set at a 7 : 3 ratio using stratified sampling, with postoperative outcome as the stratification factor. Model performance was evaluated using receiver operating characteristic curves, decision curve analysis, and F1 Score. The SHAP method was applied to analyze feature contribution. Results A total of 639 patients were included, comprising 395 males and 244 females, with a median age of 62 (55, 69) years. The adverse outcome group consisted of 191 patients, while the favorable outcome group included 448 patients, resulting in an adverse postoperative outcome incidence of 29.9%. Univariate analysis showed no significant differences between the two groups for any variables (P>0.05). Using LASSO regression, 16 feature variables were selected (including cardiopulmonary bypass support time, blood glucose on postoperative day 3, creatine kinase-MB isoenzyme, systemic inflammatory response index, etc.), and five machine learning models (GLM, RF, GBM, LightGBM, XGBoost) were constructed. Evaluation results demonstrated that the XGBoost model exhibited the best predictive performance on both the training set (n=447) and test set (n=192), with area under the curve values of 0.761 [95%CI (0.719, 0.800) ] and 0.759 [95%CI (0.692, 0.818) ], respectively. It also significantly outperformed other models in positive predictive value, and balanced accuracy in the test set. Decision curve analysis further confirmed its clinical utility across various risk thresholds. SHAP analysis indicated that variables such as cardiopulmonary bypass support time, blood glucose on postoperative day 3, creatine kinase-MB isoenzyme, and inflammatory markers (SIRI, NLR, CAR) had high contributions to the prediction. Conclusion The XGBoost model effectively predicts adverse postoperative outcomes in cardiac surgery patients. Clinically, attention should be focused on cardiopulmonary bypass support time, postoperative blood glucose control, and monitoring of inflammatory levels to improve patient prognosis.
Objective To identify genes of lipopolysaccharide (LPS) -induced acute lung injury (ALI) in mice base on bioinformatics and machine learning. Methods The acute lung injury dataset (GSE2411, GSE111241 and GSE18341) were download from the Gene Expression Database (GEO). Differential gene expression analysis was conducted. Gene ontology (GO) analysis, KEGG pathway analysis, GSEA enrichment analysis and protein-protein interaction analysis (PPI) network analysis were performed. LASSO-COX regression analysis and Support Vector Machine Expression Elimination (SVM-RFE) was utilized to identify key biomarkers. Receiver operator characteristic curve was used to evaluate the diagnostic ability. Validation was performed in GSE18341. Finally, CIBERSORT was used to analyze the composition of immune cells, and immunocorrelation analysis of biomarkers was performed. Results A total of 29 intersection DEGs were obtained after the intersection of GSE2411 and GSE111241 differentially expressed genes. Enrichment analysis showed that differential genes were mainly involved in interleukin-17, cytokine - cytokine receptor interaction, tumor necrosis factor and NOD-like receptor signaling pathways. Machine learning combined with PPI identified Gpx2 and Ifi44 were key biomarkers. Gpx2 is a marker of ferroptosis and Ifi44 is an type I interferon-induced protein, both of which are involved in immune regulation. Immunocorrelation analysis showed that Gpx2 and Ifi44 were highly correlated with Neutrophils, TH17 and M1 macrophage cells. Conclusion Gpx2 and Ifi44 have potential immunomodulatory abilities, and may be potential biomarkers for predicting and treating ALI in mince.