Heart failure with reduced ejection fraction (HFrEF) is a clinical syndrome with high morbidity and mortality. Developing cost-effective and easily deployable screening methods is crucial for improving early diagnosis and management. We propose LGC2-Net, a hierarchical fusion network that leverages multi-channel electrocardiograms (ECGs) and phonocardiograms (PCGs) for HFrEF detection. LGC2-Net simultaneously exploits complementarity across modalities and channels with its channel-specific and channel-shared branches. Each channel-specific branch employs a local-to-global hierarchical attention mechanism to capture both local and global semantic information within each ECG-PCG pair. The channel-shared branch further aligns and aggregates features from all channels, enabling effective modeling of inter-channel correlations. We established a new multi-channel ECG-PCG dataset with 2,480 synchronized recordings collected from 620 subjects using a digital stethoscope. Experiments demonstrate that LGC2-Net surpasses existing methods by 7.42% in average accuracy, highlighting its potential as an accurate, non-invasive, and scalable tool for HFrEF screening.

Myocardial infarction (MI) is a critical cardiovascular event diagnosed primarily via electrocardiogram (ECG). While many deep learning methods have been developed for automated MI detection, they often function as "black boxes," failing to incorporate the underlying physiological semantics of ECG waveforms and lacking clear interpretability. To address these challenges, we propose SEAM-Former, a segment-based attributable masking Transformer for explainable MI diagnosis. The proposed method uniquely integrates waveform segment information into its architecture to capture the structural semantics of the ECG. Furthermore, an attributable masking module is designed to generate class-specific explanations by identifying the most salient lead and waveform combinations. Evaluations on the public PTB-XL dataset demonstrate that SEAM-Former surpasses current SOTA methods, providing both superior diagnostic accuracy and explanations of high clinical coherence.

Calcific aortic valve disease (CAVD) is a serious heart valve condition with increasing global prevalence. Currently, transcatheter aortic valve implantation (TAVI) or surgical aortic valve replacement (SAVR) represents the only available treatment strategy, as no pharmaceutical therapies for CAVD are approved. The aim of this study was to identify compounds capable of inhibiting osteogenic differentiation of human aortic valve interstitial cells (hVICs), a process critically implicated in CAVD pathogenesis, and to elucidate the underlying molecular mechanism. From an in-house library of 88 compounds screened via dot-blotting, we identified chipericumin D, a natural compound extracted from Hypericum monogynum L., as a candidate exhibiting potent inhibitory activity against hVIC osteogenic differentiation. Network pharmacology analysis, molecular docking, drug affinity responsive target stability (DARTS), cellular thermal shift assay (CETSA), and surface plasmon resonance (SPR) collectively demonstrated direct binding of chipericumin D to the epidermal growth factor receptor (EGFR). Furthermore, chipericumin D suppressed activation of the EGFR/phosphatidylinositol 3-kinase (PI3K)/protein kinase B (AKT) signaling pathway in hVICs cultured under osteogenic medium (OM) conditions. These findings indicate that chipericumin D is a promising therapeutic candidate for CAVD, and provide preliminary evidence that EGFR constitutes a novel molecular target for CAVD intervention.

Aim: Atrial fibrillation and atrial flutter (AF/AFL) represent a growing public health challenge in China amid rapid population aging. This study aimed to comprehensively assess long-term trends, sex- and age-specific patterns, driving factors, and future projections of AF/AFL burden in China from 1990 to 2023. Methods: Data were obtained from Global Burden of Disease 2023. Temporal trends, driving factors, and future projections were evaluated using joinpoint regression, age-period-cohort models, decomposition analysis, frontier analysis, and autoregressive integrated moving average (ARIMA) models. Results: From 1990 to 2023, the absolute numbers of AF/AFL cases, deaths, and disability-adjusted life years (DALYs) increased substantially in China, whereas age-standardized mortality and DALY rates declined overall, with a recent upturn after 2020. Incidence and prevalence were generally higher in males, whereas females had higher mortality and DALYs at older ages. Population aging was the dominant contributor to increases in incidence, prevalence, mortality, and DALYs. Age-period-cohort models analyses showed that among individuals born after 1944, the burden of AF/AFL was higher in males than in females. Frontier analysis indicated that China still lags behind several high Socio-demographic Index regions in AF/AFL burden control. ARIMA projections suggested declining mortality and DALYs but heterogeneous future trends in incidence and prevalence by sex. Conclusions: Despite improvements in age-standardized mortality and DALYs, the overall burden of AF/AFL in China continues to increase, primarily driven by population aging. Targeted prevention, early detection, and optimized management strategies - particularly among older adults and high-risk males - are urgently needed.

Background and Objective:The field of surgical treatment for valvular heart disease (VHD) has progressed rapidly in 2025. This review aims to summarize the year's key research, focusing on comparisons of mainstream techniques for aortic, mitral, and tricuspid valve diseases, innovations in repair techniques, and advances in emerging technologies, to provide evidence-based support for individualized clinical decision-making. Methods:A structured search of PubMed database was performed to identify randomized controlled trials (RCTs), registry studies, and meta-analyses published in 2025, and the evidence was synthesized narratively. Key Content and Findings:In the aortic valve arena, long-term follow-up from the PARTNER 3 and Evolut Low Risk trials confirmed the non-inferiority of transcatheter aortic valve replacement (TAVR) in low-risk patients. However, real-world data suggested higher long-term risks with TAVR in patients with bicuspid aortic valves (BAV) and younger patients, whereas women demonstrated greater benefit. The indication for early intervention in asymptomatic severe aortic stenosis (AS) was reinforced. Aortic valve repair and the Ross procedure accumulated more robust evidence for long-term survival and quality of life in younger patients. In the mitral valve domain, the MITRACURE study highlighted gaps between real-world practice and guidelines, including delayed referral and suboptimal repair rates for mitral regurgitation (MR). The advantages of repair for degenerative disease were further solidified, though techniques for posterior leaflet prolapse and management strategies for atrial functional regurgitation (AFMR) require optimization. Transcatheter edge-to-edge repair (TEER) was increasingly used in high-risk patients but yielded inferior long-term outcomes compared to surgery, and surgical rescue after failed TEER carried high risk. For the tricuspid valve, comparative effectiveness of transcatheter vs. surgical repair and the timing of intervention emerged as key research foci. Emerging technologies such as polymer valves, partial heart transplantation, and artificial intelligence (AI)-assisted analysis demonstrated preliminary potential. Conclusions:The 2025 evidence reinforces a paradigm shift from risk-score-based decision-making toward individualized valve care. Transcatheter techniques have expanded their indications, but surgery remains irreplaceable in young, low-risk, and BAV patients, as well as in degenerative MR. Real-world gaps in guideline adherence, delayed referral, and suboptimal repair rates require urgent attention. Emerging technologies show early promise, although their long term durability and clinical value await further validation.

Hypoglycemia is a major barrier to safe diabetes management. Although deep learning has been widely applied to blood glucose (BG) prediction, most studies provide limited hypoglycemia forewarning and are trained on small type 1 diabetes cohorts with restricted generalizability. We developed MT-HypoNet, a multitask neural network for real-time BG prediction and hypoglycemia forewarning from continuous glucose monitoring data. To improve detection near the hypoglycemia boundary, we introduce a statistically guided soft-label strategy. MT-HypoNet was validated on a multicenter cohort of 1,662 patients with type 1 and type 2 diabetes and prospectively evaluated in 36 perioperative patients with type 2 diabetes. In internal validation, MT-HypoNet achieved an AUC of 0.946 (95% CI: 0.946-0.947) for hypoglycemia forewarning and an RMSE of 19.84 ± 4.92 mg/dL for BG prediction. It generalized well to external datasets and maintained high prospective performance (AUC 0.966; RMSE 16.62 ± 4.01 mg/dL), supporting proactive management and improved safety.

Digital subtraction angiography (DSA) devices guide procedures across numerous diseases, performed on more than 100,000 patients daily worldwide. However, these procedures expose patients and healthcare providers to radiation, increasing the risk of health issues. Despite many low-dose DSA imaging methods proposed, none have been prospectively clinically validated. In this study, 46,829 patients (over 5 million DSA images) from 70 centers were used to iterate our previously developed generative artificial intelligence system (named GenDSA-V2). A total of 1,068 patients (533 in intervention arm and 535 in control arm), with suspected cerebral aneurysms (n = 435), lung cancer (n = 417) or advanced liver cancer (n = 216), meeting surgical criteria, were enrolled to validate the GenDSA-V2. The primary outcome was radiation dose, while secondary outcomes included efficiency, operation time and intraoperative complications. Group assignments were blinded to patients, surgeons and investigators, while technicians were aware but not involved in data collection or analysis. The GenDSA-V2 group showed substantially reduced radiation exposure, with an air kerma (AK) of 151.3 ± 125.1 mGy compared to 457.4 ± 407.4 mGy in the standard clinical protocols (SCP) group (mean difference = -306.1 mGy, 95% confidence interval (CI) = -342.3 to -269.9, P < 0.001 for superiority) and a dose-area product (DAP) of 4009.7 ± 2767.9 μGy m2 versus 12531.6 ± 9145.9 μGy m2 (mean difference = -8521.9 μGy m2, 95% CI = -9333.1 to -7710.7, P < 0.001 for superiority). Mean operation time was 33.1 ± 10.8 min in the SCP group and 34.8 ± 11.8 min in the GenDSA-V2 group (mean difference = 1.7 min, 95% CI = 0.3 to 3.1, P < 0.001 for noninferiority). Complication rates were similar (SCP = 8.1%, GenDSA-V2 = 7.5%, mean difference = -0.6%, 95% CI = -3.8% to 2.6%, P < 0.001 for noninferiority). The GenDSA system reduces radiation exposure to both physicians and patients by approximately two-thirds during DSA-guided procedures, demonstrating substantial clinical and translational value. Chinese Clinical Trial Registry: ChiCTR2400084789 .

Automatic analysis methods of electrocardiograms (ECGs) usually required large-scale annotated training data, but the annotation process is extremely time-consuming. While semi-supervised learning can leverage unlabeled data, its performance depends heavily on the quality of the initial labeled subset. Active learning has been used to identify the most informative samples for annotation, but conventional approaches face three critical limitations: (1) dependency on manual intervention for iterative query design, (2) prohibitive computational costs during sample selection, and (3) limited compatibility with semi-supervised learning frameworks. To address these limitations, we proposed an Unsupervised Active Feature-selective Semi-Supervised Learning (UAFSSL) framework for ECG analysis, including an unsupervised feature selection-based active learning module and a semi-supervised learning module. UAFSSL captures latent data distributions via unsupervised feature extraction, selects diverse and representative samples using pseudo-label clustering, and integrates seamlessly with semi-supervised learning to eliminate human intervention. We validated our algorithm on an ECG waveform segmentation task and an atrial fibrillation detection task. In the waveform segmentation task, our method improved the F1-score for P-wave delineation by 2.4% compared to random sampling, using only 5% of labeled samples. For the atrial fibrillation detection task, we evaluated our method on both the AFDB and a 24-hour dataset collected from 500 atrial fibrillation patients. Using only 200 labeled samples for model training, our method achieved AUC improvements of 2.5% and 2.2% over random sampling in five-fold cross validation. This is the first study to integrate unsupervised active learning with semi-supervised learning for automatic ECG analysis, offering a robust, automated solution to reduce annotation costs while enhancing clinical applicability.

Calcific aortic valve disease (CAVD), the most common human valve disease on a global scale, ranks and persists as an unaddressed clinical challenge. This is primarily attributed to the absence of efficacious pharmacological approaches. The Nuclear Receptor Subfamily 4 Group A Member 1 (NR4A1), intricately associated with the pathogenesis of multiple cardiovascular diseases, has emerged as a pivotal target for the diagnosis and treatment of numerous ailments. However, the specific molecular mechanisms and the functional significance of NR4A1 in the pathogenesis of CAVD are yet to be comprehensively elucidated. By performing in-depth analyses on human aortic valve tissues and carrying out functional investigations using primary valvular interstitial cells (VICs), we were able to demonstrate that NR4A1 significantly facilitated cellular proliferation and intensifies the osteogenic differentiation process of VICs. Evidently, this is reflected in the elevated expression of key osteogenic markers, namely runt-related transcription factor 2 (RUNX2) and alkaline phosphatase (ALP). Mechanistically, the pro-calcific effects were achieved via NR4A1-dependent modulation of the cell cycle regulatory protein Cyclin D2 (CCND2). Significantly, in vivo investigations employing ApoE-/- mice maintained on a high-fat Western diet demonstrated that pharmacological suppression of NR4A1 efficiently mitigated the advancement of aortic valve calcification. These discoveries not merely determine NR4A1 to be a crucial modulator in cellular proliferation, thereby accelerating valvular calcification, but also present compelling evidence advocating for targeting NR4A1 may represent a potential therapeutic strategy for CAVD.

BACKGROUND:Given the biases and ethical concerns of AI models, the fully automatic diagnosis of diseases in clinical settings is challenging. In contrast, clinician-AI collaboration is considered essential to ensure the validity and reliability of utilizing AI models in clinical practice. However, effective strategies for clinician-AI collaboration remain largely unexplored. METHODS:This study proposed a three-step general clinician-AI collaboration pipeline aimed at improving disease diagnosis efficiency: first, utilizing large real-world clinical datasets to evaluate and clarify clinicians' diagnostic strengths/weaknesses; second, developing an AI model to complement clinicians' weakness in disease diagnosis; and finally, proposing a clinician-AI collaboration strategy to leverage the strengths of both AI and clinicians. The effectiveness of this pipeline was validated through a study focusing on clinical paroxysmal atrial fibrillation (PAF) detection, utilizing 24-h Holter recordings from over 30,000 patients. FINDINGS:In PAF detection, clinicians alone required a significant amount of time to identify the data and still overlooked 13.7% of PAF patients but successfully identified all non-atrial fibrillation (AF) patients. Conversely, AI alone rarely missed PAF patients but misidentified 23.3% of non-AF patients as having PAF. After implementing the proposed clinician-AI collaboration strategy, all patients were correctly identified, and clinicians' workload was reduced by 76.7%. CONCLUSIONS:This study improves both the efficiency and reliability of PAF detection, bridging the gap between AI model development and its clinical application, thereby effectively promoting the application of AI models in clinical AF screening. FUNDING:This study was supported in part by the National Natural Science Foundation of China.

AIMS:Calcific aortic valve disease (CAVD) is becoming more prevalent with the population ageing; however, there is currently no medical therapy available. During early lipid deposition, low-density lipoprotein (LDL) mediates chronic inflammation and accelerates calcification progression. However, the mechanism still needs to be further explored. METHODS AND RESULTS:The study identified the transcription factor FOXS in human valvular interstitial cells (VICs) as a pivotal regulator in aortic valve calcification. Bulk RNA-seq and qRT-PCR analysis were conducted to establish that FOXS1 is induced by oxidized LDL (oxLDL) in VICs. To elucidate the role of FOXS1 in osteogenic differentiation, small interfering RNA and recombinant adenovirus were utilized to modulate FOXS1 expression in VICs. High-fat diet (HFD)-fed Apoe-/-Foxs1-/- mice served as an in vivo model to investigate the role of FOXS1 in aortic valve calcification. Analysis from bulk RNA-seq, qRT-PCR, and western blot indicated significant activation of FOXS1 by oxLDL in VICs, with silencing of FOXS1 inhibiting oxLDL-induced osteogenic differentiation. Deletion of FOXS1 markedly reduced aortic valve calcification in HFD-fed Apoe-/- mice, as shown by decreased calcium deposition in the aortic valve leaflets. RNA-seq and chromatin immunoprecipitation sequencing were performed to reveal the regulatory mechanisms of FOXS1, uncovering direct interactions with the promoter of BSCL2, which subsequently inhibits the expression of ABCA1 and ABCG1 via the PPARγ/LXRα axis. The study demonstrated that FOXS1 mediates VICs' cholesterol transport dysfunction through BSCL2, ABCA1, and ABCG1 using Bodipy-cholesterol and showed that intracellular cholesterol accumulation can activate the NLRP3 inflammasome, promoting osteogenic differentiation of VICs. Additionally, it was found that IMM-H007 and recombinant BSCL2 could reduce aortic valve calcification both in vitro and in vivo. CONCLUSION:We identified that an oxLDL-induced transcription factor FOXS1 inhibits ABCA1 and ABCG1 expression via the BSCL2/PPARγ/LXRα axis and promotes cholesterol transport dysfunction and the activation of NLRP3 inflammasome in VICs, thereby accelerating the progression of CAVD.

Background: The causal relationship between migraines and patent foramen ovale (PFO) remains controversial, and a major unresolved question is how to define migraines attributable to PFO. Thus, this study aimed to determine if brain lesions could be a potential indicator of PFO-related migraines. Methods: Consecutive migraine patients from 2017 to 2019 who underwent transthoracic echocardiography or transcranial Doppler examination with an agitated saline contrast injection were assessed for right-to-left shunts. We then presented diffusion-weighted imaging (DWI) in brain magnetic resonance imaging and its association with PFO in the included patients. Results: A total of 424 patients with a mean age of 44.39 ± 12.06 years were included in this retrospective study. Among them, 244 patients (57.5%) had PFO, and 246 patients (58%) had subclinical brain lesions—the brain lesions presented as single or multiple scattered lesions. No association was observed between PFO prevalence and brain lesions in the total cohort (odds ratio (OR) 0.499); however, a significant association was observed in patients aged less than 46 years (OR, 3.614 in the group aged <34 years, 95% confidence interval (CI) 1.128–11.580, and 3.132 in the group of 34 years ≤ age < 46 years, 95% CI 1.334–7.350, respectively). Lesions in patients with PFO observed using DWI came more from the anterior or multiple than the posterior vascular territory (p = 0.033). DWI lesion numbers, location, and right-to-left shunt amounts did not affect the association between DWI-observed lesions and PFO. Conclusions: This study demonstrated that subclinical brain lesions are associated with PFO and may be used as a potential predictor of PFO-related migraines in patients aged less than 46 years. This may help identify candidate patients for PFO closure in future clinical decisions.

Accurate electrocardiogram (ECG) segmentation is critical for diagnosing and monitoring cardiac conditions. However, the accuracy of ECG segmentation across different heart rhythm types remains a challenge, and its practical utility in disease diagnosis remains to be fully validated. To address these challenges, we propose Y-Net, a deep learning model designed to perform robust ECG segmentation under both single-lead and multi-lead input modes. The model incorporates a dual-branch structure and a two-stage training strategy to ensure adaptability across various clinical scenarios. We evaluated Y-Net on two 12-lead ECG segmentation datasets: LUDB, a public dataset, and RDB, a privately annotated dataset based on public data but annotated specifically by our team. YNet demonstrated robust performance across datasets and rhythm types, achieving F1 scores of 99.60% and 99.44% in intra-dataset evaluations, and 99.03% and 98.24% in inter-dataset tests. To improve interpretability, we introduce an intermediate feature visualization method and apply segmentation results directly to atrial fibrillation (AF) detection based on P-wave absence. This morphology-based approach achieves AUCs of 0.946, 0.971, and 0.983 on the PhysioNet2017, CPSC2018, and AFDB datasets, respectively, without the need for additional classifiers. These results highlight the effectiveness and clinical potential of Y-Net as a transparent and adaptable tool for ECG segmentation and interpretation across diverse cardiac rhythms.

Automatic analysis of electrocardiogram signals has been widely applied in the intelligent detection of cardiovascular diseases, however, class imbalance caused by rare disease samples limits model performance. In this paper, we propose a progressive generation strategy for synthesizing ECG signals, consisting of two stages: from structural skeletons to high-fidelity detail refinement. Specifically, we first train a generator to produce structural representations of ECG signals in the form of square-wave encodings, which capture coarse patterns. Then, an attention-based refinement module is introduced to fuse the detailed features from real ECG signals with the coarse features of the square-wave representation. This design ensures stable training and high-quality signal generation. Using the Resting ECG Segmentation Dataset, we synthesize AFIB, AT, and AF signals. Our method outperforms SOTA models with the lowest ED (12.815) and KLD (0.325) in feature space, and achieves an F1-score of 0.672-4% higher than the best baseline (0.632).

Calcific aortic valve disease (CAVD) is prevalent in developed nations and has emerged as a pressing global public health concern due to population aging. The precise etiology of this disease remains uncertain, and recent research has primarily focused on examining the role of valvular interstitial cells (VICs) in the development of CAVD. The predominant treatment options currently available involve open surgery and minimally invasive interventional surgery, with no efficacious pharmacological treatment. This article seeks to provide a comprehensive understanding of valvular endothelial cells (VECs) from the aspects of valvular endothelium-derived nitric oxide (NO), valvular endothelial mechanotransduction, valvular endothelial injury, valvular endothelial-mesenchymal transition (EndMT), and valvular neovascularization, which have received less attention, and aims to establish their role and interaction with VICs in CAVD. The ultimate goal is to provide new perspectives for the investigation of non-invasive treatment options for this disease.

The prevalence of calcific aortic valve disease (CAVD) remains substantial while there is currently no medical therapy available. Forkhead box O1 (FOXO1) is known to be involved in the pathogenesis of cardiovascular diseases, including vascular calcification and atherosclerosis; however, its specific role in calcific aortic valve disease remains to be elucidated. In this study, we identified FOXO1 significantly down-regulated in the aortic valve interstitial cells (VICs) of calcified aortic valves by investigating clinical specimens and GEO database analysis. FOXO1 silencing or inhibition promoted VICs osteogenic differentiation in vitro and aortic valve calcification in Apoe-/- mice, respectively. We identified that FOXO1 facilitated the ubiquitination and degradation of RUNX2, which process was mainly mediated by SMAD-specific E3 ubiquitin ligase 2 (SMURF2). Our discoveries unveil a heretofore unacknowledged mechanism involving the FOXO1/SMURF2/RUNX2 axis in CAVD, thereby proposing the potential therapeutic utility of FOXO1 or SMURF2 as viable strategies to impede the progression of CAVD.

The association between malnutrition and outcomes of heart transplantation (HTx) has not been well studied. The purpose of this article was to evaluate the prognostic value of three different nutrition indices in HTx, including CONUT (Controlling Nutritional Status), NRI (Nutritional Risk Index) and GNRI (Geriatric Nutritional Risk Index). A total of 438 patients who underwent THx from January 2015 to December 2020 were included in this study. The nutritional status of the patients was evaluated by CONUT, NRI and GNRI. Kaplan-Meier (KM) curves were constructed to compare the difference in overall survival (OS) between the normal and malnutrition groups in each index. Cox regression analysis was used to identify the independent risk factors of OS. The predictive power was compared by time-dependent ROC and time-dependent ccurves. Logistic regression model was used to evaluate the relationship between these three nutrition indices and postoperative clinical events. 336 (76.7

BACKGROUND:Early diagnosis of atrial fibrillation (AF) is important for preventing stroke and other complications. Predicting AF risk in advance can improve early diagnostic efficiency. Deep learning has been used for disease risk prediction; however, it lacks adherence to evidence-based medicine standards. Identifying the underlying mechanisms behind disease risk prediction is important and required. METHODS:We developed an explainable deep learning model called HBBI-AI to predict AF risk using only heart beat-to-beat intervals (HBBIs) during sinus rhythm. We proposed a possible AF mechanism based on the model's explainability and verified this conjecture using confirmed AF risk factors while also examining new AF risk factors. Finally, we investigated the changes in clinicians' ability to predict AF risk using only HBBIs before and after learning the model's explainability. FINDINGS:HBBI-AI consistently performed well across large in-house and external public datasets. HBBIs with large changes or extreme stability were critical predictors for increased AF risk, and the underlying cause was autonomic imbalance. We verified various AF risk factors and discovered that autonomic imbalance was associated with all these factors. Finally, cardiologists effectively understood and learned from these findings to improve their abilities in AF risk prediction. CONCLUSIONS:HBBI-AI effectively predicted AF risk using only HBBI information through evaluating autonomic imbalance. Autonomic imbalance may play an important role in many risk factors of AF rather than in a limited number of risk factors. FUNDING:This study was supported in part by the National Key R&D Program and the National Natural Science Foundation of China.

IntroductionCalcific aortic valve disease (CAVD) is increasingly prevalent among the aging population, and there is a notable lack of drug therapies. Consequently, identifying novel drug targets will be of utmost importance. Given that type 2 diabetes is an important risk factor for CAVD, we identified key genes associated with diabetes - related CAVD via various bioinformatics methods, which provide further potential molecular targets for CAVD with diabetes.MethodsThree transcriptome datasets related to CAVD and two related to diabetes were retrieved from the Gene Expression Omnibus (GEO) database. To distinguish key genes, differential expression analysis with the “Limma” package and WGCNA was applied. Machine learning (ML) algorithms were employed to screen potential biomarkers. The receiver operating characteristic curve (ROC) and nomogram were then constructed. The CIBERSORT algorithm was utilized to investigate immune cell infiltration in CAVD. Lastly, the association between the hub genes and 22 types of infiltrating immune cells was evaluated.ResultsBy intersecting the results of the “Limma” and WGCNA analyses, 727 and 190 CAVD - related genes identified from the GSE76717 and GSE153555 datasets were obtained. Then, through differential analysis and interaction, 619 genes shared by the two diabetes mellitus datasets were acquired. Next, we intersected the differential genes and module genes of CAVD with the differential genes of diabetes, and the obtained genes were used for subsequent analysis. ML algorithms and the PPI network yielded a total of 12 genes, 10 of which showed a higher diagnostic value. Immune cell infiltration analysis revealed that immune dysregulation was closely linked to CAVD progression. Experimentally, we have verified the gene expression differences of MFAP5, which has the potential to serve as a diagnostic biomarker for CAVD.ConclusionIn this study, a multi-omics approach was used to identify 10 CAVD-related biomarkers (COL5A1, COL5A2, THBS2, MFAP5, BTG2, COL1A1, COL1A2, MXRA5, LUM, CD34) and to develop an exploratory risk model. Western blot (WB) and immunofluorescence experiments revealed that MFAP5 plays a crucial role in the progression of CAVD in the context of diabetes, offering new insights into the disease mechanism.