Unknown Type Research Articles

An adaptive thresholding approach for open set fault diagnosis

Abstract Cross-domain fault diagnosis using deep learning plays a critical role in ensuring the reliability and safety of mechanical systems. However, real-world industrial scenarios often involve unknown fault classes, which introduce significant challenges beyond environmental differences between training and testing phases. These unknown fault classes, which do not appear in the training data, create a cross-domain open set fault diagnosis problem where the target domain includes both known and unknown fault types with distinct distribution characteristics. Traditional domain adaptation methods that align source and target domains often overlook the spatial distribution of each class in the feature space, leading to potential negative transfer and misclassification of unknown faults. To address these challenges, this paper proposes a k-nearest neighbors based adaptive thresholding (KNNAT) method, which dynamically adjusts classification thresholds based on the spatial distribution of each class in the feature space. This approach effectively isolates unknown faults, reducing their impact on domain adaptation and improving the reliability of the diagnostic process. Extensive experiments on the publicly available CWRU bearing and PHM09 datasets demonstrate that the proposed KNNAT method outperforms other state-of-the-art methods, achieving higher accuracy and robustness in identifying known faults while successfully isolating unknown faults. These results highlight the potential of using the KNNAT method to enhance the reliability of mechanical systems in cross-domain fault diagnosis applications.

Comparative retrospective study of injuries during military parachuting training and airborne operations from 2013 to 2020

IntroductionFrench paratroopers have been deployed during airborne operations for nearly a century. Parachute operations have increased since the start of Operations Serval and Barkhane in the Sahelo-Saharan strip. These military...

K-means Optimized Network Security Management in the Internet Context

The rapid development of the Internet has had a broad and profound impact on humanity, making information acquisition and dissemination more convenient. It has also brought significant opportunities and benefits to business and the economy. However, there are some issues, such as personal factors and data security concerns. In order to solve the above problems, the K-means algorithm is optimized from the perspectives of K-value validity index, feature weighting and three-branch decision making. First, the optimal clustering results are determined according to K-value validity index, and the influence of different dimensional features on clustering is considered for feature weighting, and the uncertain objects in the three-branch decision deadlock class are divided into the boundary domain. The delay decision of the boundary domain data is carried out, and the K-means clustering optimization algorithm is improved by combining the above three aspects, and the intelligent network security management system is developed on this basis. The results showed that the K-means optimization algorithm achieved the highest average accuracy rate, adjusted Morandi index, and adjusted mutual information across various datasets, with values of 96.01%, 0.866, and 0.869, respectively. In practical network attack scenarios, the K-means optimization algorithm attained an attack threat recognition accuracy of 94.38%. Under unknown network attack types, its detection rate and false alarm rate were 94.63% and 1.32%, respectively. Surveys conducted post-implementation of the intelligent network security management system indicated that over 90% of users were satisfied with their experience of the system. In summary, the proposed method accurately identifies potential network threats in network data, fulfilling performance requirements for network security management systems and ensuring the security of network resources.

Comparing conventional and phytoscreening methods to detect subsurface chemical contaminants: A test case of volatile organic compounds in an urban setting

The nationwide prevalence of brownfields, with often unknown types and quantities of subsurface chemical contaminants, highlights the need for rapid, cost-effective, and noninvasive methods to reduce routes of exposure. In post-industrial cities such as Detroit, Michigan, anthropogenic volatile organic compounds (VOCs), known to negatively impact human health, are typically detected at brownfields through conventional methods, e.g. screening soil, and groundwater. Recently, the method of phytoscreening–the chemical analysis of plant tissues to provide evidence for belowground contamination–has become a viable alternative to conventional methods. However, few studies have been designed to directly compare conventional and plant-based methods of detecting VOCs. To fill this knowledge gap, we sampled and compared the concentration of six VOCs including BTEX, PCE, and TCE detected in conventional media (soil, soil vapor, groundwater, sewer vapor) and different plant tissue (tree core, leaf, root, shoot) at two brownfields sites in Detroit: an abandoned gas station with a leaking underground storage tank, and a former dry cleaning facility. Our results suggest that the concentrations of VOCs detected in plants are similar to or in some cases greater than conventional methods and can differ across the growing season. For example, leaves and roots detected, on average, a higher concentration of VOCs compared to shoots and tree cores, however, TCE and PCE were generally in higher concentrations in soil and soil vapor. Moreover, the frequency at which conventional versus phytoscreening methods failed to detect VOCs was similar at one site and higher at another, suggesting that phytoscreening may yield fewer non-detects at known sites of contamination. While additional work is needed to understand the relationship between concentrations of VOCs detected in soil versus co-located plant samples, our results suggest that phytoscreening may be a viable and reliable method to detect belowground chemical contaminants while reducing screening times and cost, and increasing access to private property.

Disseminated Bacille Calmette-Guérin Vaccine-Induced Disease in a Sample of Iranian Children: A Longitudinal Case-Series Study

Background: Approximately a quarter-million Iranian newborns receive the Bacille Calmette-Guérin (BCG) vaccine annually. However, in immunodeficient infants, this vaccine can lead to disseminated BCG disease (DBD). Objectives: This study aims to determine the clinical presentation, comorbidities, underlying immunodeficiency, and prognosis of DBD in a group of Iranian children. Methods: In a longitudinal case-series study, data were collected from the files of infants and children diagnosed with DBD from 2005 to 2017. Immunodeficiency screening was conducted for each patient. For children with normal immunodeficiency screening results, further testing for Mendelian susceptibility to mycobacterial diseases (MSMD) was performed. Detection of DBD in children was achieved by evaluating gastric lavage and bone marrow aspiration samples for mycobacterium. Results: Most of the 22 patients were immunocompromised, with the following distribution: 31.8% had severe combined immunodeficiency (SCID), 45.5% had MSMD (specifically IL-12Rβ1 deficiency), one patient (4.5%) had Wiskott-Aldrich syndrome (WAS), and the remaining 18.1% had unknown immunodeficiency types. Most patients with MSMD were successfully treated and did not show relapse during the follow-up period, even after discontinuing anti-tuberculosis (TB) medications. Conclusions: Due to the similarity of its manifestations to sepsis, diagnosing systemic infections caused by the BCG vaccine in children requires a high level of clinical suspicion and appropriate diagnostic measures, such as mycobacterial culture, biochemical speciation, or polymerase chain reaction (PCR). These diagnostic steps should be taken promptly in cases of DBD, with concurrent treatment using anti-tuberculosis drugs and, if possible, targeted therapies for underlying immunodeficiency.

Pregnancy planning in women with epilepsy: A single center observational study with focus on epilepsy type

PurposeTo explore various aspects of pregnancy planning in women with epilepsy and to identify factors needing particular attention in the counselling of these patients with focus on epilepsy type. Methods285 pregnancies in 192 women were collected from the EURAP registry in Trondheim, Norway. Medical records were reviewed to validate diagnoses and types of epilepsy according to revised ILAE classifications. ResultsTen women proved to have non-epileptic conditions, leaving 274 pregnancies in 182 patients for inclusion. In 40 %, the epilepsy was focal, in 45 % generalized, including 18 % with JME. In 14 %, the epilepsy type was unknown. Pregnancies were planned in 64 %; 16 % were unintended and 20 % undetermined. Unintended pregnancies occurred in 15 % with focal and in 17 % with generalized epilepsy and in only 10 % of the JME subgroup. Planned pregnancy was associated with both preconception folic acid intake (p < 0.001) and breastfeeding ≥6 months (p = 0.011). Epilepsy of unknown type had the lowest rates of intended pregnancy and folic acid use. ConclusionWe found no difference in pregnancy planning between focal and generalized epilepsy. Intended pregnancy was strongly associated with both folic acid and breastfeeding. The JME subgroup did not perform worse but rather above average regarding family planning and breastfeeding. The lowest proportion of folic acid intake was found in epilepsy of unknown type in which seizure control is common, and patients may receive less attention from the specialist health service. Appropriate counselling regarding pregnancy should reach out to all fertile women regardless of epilepsy type and seizure control.

Diagnosis and treatment of magnetic foreign bodies in the gastrointestinal tract: the experience of four centers

BACKGROUND: Magnetic foreign bodies in the gastrointestinal tract is a relatively rare problem in urgent pediatric surgery. Meanwhile, among cases with foreign bodies, they are the most difficult both in terms of timely diagnostics and timely treatment. There are no generally accepted therapeutic and diagnostic standards, and traditional algorithms of patient management are focused on the dynamic observation and are not correct. More so, algorithms described in the foreign literature contradict each other. The rate of surgical interventions and the risk of complications are high. Surgical tactics has many "white spots". All these issues remain open for discussion. AIM: To analyze the experience of four medical centers in the diagnostics and treatment of patients with magnetic foreign bodies in the gastrointestinal tract. METHODS: A retrospective analysis of diagnostic and treatment techniques applied in patients with magnetic foreign bodies in their gastrointestinal tract from four pediatric surgical hospitals in 2010–2023 was made. In the study there were patients with neodymium magnetic balls from the constructor (38 patients, 50%), neodymium magnets of other shapes and for other purposes (30 patients, 39.5%), ferrite magnets (6 patients, 7.9%), unknown types of magnets (2 patients, 2.6%). RESULTS: Patients’ average age was (4±3.65) years. The fact of foreign body swallowing was revealed in anamnesis of 46 (60.5%) patients. Mental illnesses were diagnosed in 4 children — 5.3%, which is higher than the average incidence in pediatric population. Clinical symptoms were registered in 4 (11.8%) children with a single magnet foreign body and in 33 (78.6%) children with multiple ones. The basic diagnostic technique was an overview radiography of the abdominal cavity (70 cases, 92%), which allowed to identify both a foreign body and complications (signs of intestinal obstruction, free gas in the abdominal cavity). Esophagogastroduodenoscopy was performed in 50 cases (65.8%), out of which one foreign body was detected in 34 (68%) cases, and in 28 patients, foreign bodies were removed. Other diagnostic methods (ultrasound, spiral computed tomography) were used much less frequently. Among curative techniques, open surgeries prevailed — 32 cases (42%). Laparoscopy was used in few cases. CONCLUSION: The timely diagnostics of magnetic foreign bodies in gastrointestinal tract often depends on anamnesis findings, and if there is no such, diagnostics most often is delayed and is made only intraoperatively. Among the diagnostic techniques, it is difficult to identify a universal one that can be recommended as the "gold standard"; however, the most widely used and well-informative is abdominal X-ray and esophagogastroduodenoscopy. As to surgical tactics, the role of laparoscopic interventions is relatively low. Most often, endoscopy or "open" surgery are performed.

Identification of atypical pediatric diabetes mellitus cases using electronic medical records

IntroductionThere are no established methods to identify children with atypical diabetes for further study. We aimed to develop strategies to systematically ascertain cases of atypical pediatric diabetes using electronic medical...

SOAMC: A Semi-Supervised Open-Set Recognition Algorithm for Automatic Modulation Classification

Traditional automatic modulation classification methods operate under the closed-set assumption, which proves to be impractical in real-world scenarios due to the diverse nature of wireless technologies and the dynamic characteristics of wireless propagation environments. Open-set environments introduce substantial technical challenges, particularly in terms of detection effectiveness and computational complexity. To address the limitations of modulation classification and recognition in open-set scenarios, this paper proposes a semi-supervised open-set recognition approach, termed SOAMC (Semi-Supervised Open-Set Automatic Modulation Classification). The primary objective of SOAMC is to accurately classify unknown modulation types, even when only a limited subset of samples is manually labeled. The proposed method consists of three key stages: (1) A signal recognition pre-training model is constructed using data augmentation and adaptive techniques to enhance robustness. (2) Feature extraction and embedding are performed via a specialized extraction network. (3) Label propagation is executed using a graph convolutional neural network (GCN) to efficiently annotate the unlabeled signal samples. Experimental results demonstrate that SOAMC significantly improves classification accuracy, particularly in challenging scenarios with limited amounts of labeled data and high signal similarity. These findings are critical for the practical identification of complex and diverse modulation signals in real-world wireless communication systems.

Single cell transcriptional analysis of human adenoids identifies molecular features of airway microfold cells.

The nasal, oropharyngeal, and bronchial mucosa are primary contact points for airborne pathogens like Mycobacterium tuberculosis (Mtb), SARS-CoV-2, and influenza virus. While mucosal surfaces can function as both entry points and barriers to infection, mucosa-associated lymphoid tissues (MALT) facilitate early immune responses to mucosal antigens. MALT contains a variety of specialized epithelial cells, including a rare cell type called a microfold cell (M cell) that functions to transport apical antigens to basolateral antigen-presenting cells, a crucial step in the initiation of mucosal immunity. M cells have been extensively characterized in the gastrointestinal (GI) tract in murine and human models. However, the precise development and functions of human airway M cells is unknown. Here, using single-nucleus RNA sequencing (snRNA-seq), we generated an atlas of cells from the human adenoid and identified 16 unique cell types representing basal, club, hillock, and hematopoietic lineages, defined their developmental trajectories, and determined cell-cell relationships. Using trajectory analysis, we found that human airway M cells develop from progenitor club cells and express a gene signature distinct from intestinal M cells. Surprisingly, we also identified a heretofore unknown epithelial cell type demonstrating a robust interferon-stimulated gene signature. Our analysis of human adenoid cells enhances our understanding of mucosal immune responses and the role of M cells in airway immunity. This work also provides a resource for understanding early interactions of pathogens with airway mucosa and a platform for development of mucosal vaccines.

Cardiovascular health metrics and all-cause mortality in osteoarthritis, inflammatory arthritis, and unclassified arthritis patients: a national prospective cohort study

BackgroundArthritis notably elevates mortality risk. It remains unclear whether the cardiovascular health (CVH) metrics improves the risk of all-cause mortality in patients with all types of arthritis.MethodsThis study data from the National Health and Nutrition Examination Survey to probe the link between CVH and all-cause mortality among arthritis sufferers in the United States. CVH evaluation employed the Life's Essential 8 metrics. Mortality outcomes were scrutinized using Cox proportional hazard regression models. Additionally, a restricted cubic spline analysis delineated the linear relationship between CVH and mortality. The study also delved into the singular impact of each CVH component on mortality.ResultsIn the cohort of 5919 patients with arthritis, improved CVH was linked to lower all-cause mortality. Specifically, each 10-point increment in CVH score was associated with a substantial decline in all-cause mortality risk [unadjusted hazard ratio (HR): 0.77, 95% Confidence Interval (95% CI): 0.71–0.83, P < 0.001]. Adjustments for age, sex, race, and social determinants of health further refined the HR to 0.72 (95% CI: 0.67–0.79, P < 0.001). Higher versus lower CVH scores at baseline markedly reduced mortality risk, with the most substantial decrease seen in those with ideal CVH metrics (HR: 0.39, 95% CI: 0.26–0.59, P < 0.001). Similar results were not observed in patients with inflammatory arthritis, but were seen in those with osteoarthritis or degenerative arthritis, and unknown types of arthritis.ConclusionIdeal CVH substantially decreases all-cause mortality risk among patients with arthritis, confirming the critical role of CVH in arthritis management. This study advocates for CVH interventions as part of comprehensive arthritis treatment plans.

Electric vehicle energy consumption prediction for unknown route types using deep neural networks by combining static and dynamic data

Accurate energy consumption prediction of electric vehicles (EVs) is crucial for drivers considering long trips. All the data should be provided beforehand to determine the energy consumption at the beginning of the trip. Although dynamic vehicle data (vehicle speed, state-of-charge, acceleration, etc.) cannot be known before the trip, factors related to the specified route (route type, elevation, average speed, weather, driving time, etc.) can be used to predict the consumed energy. These factors can be categorized as static and dynamic features, and thus, the question of how to effectively use static and dynamic data arises. This paper investigates the problem of predicting the energy consumption of an EV for a predetermined trip using a deep neural network (DNN) model that effectively uses static features along with dynamic segment features. Furthermore, we address the problem where the route types are unknown in advance. To include more information in the prediction model, we clustered the speed profiles using shape-based clustering with dynamic time warping (DTW) to predict the route type and used the cluster labels as static inputs. Real driving data collected from various drivers of a specific EV were used to train the DNN. The proposed DNN model was compared with the average energy consumption (AEC) model and five machine learning models. The results show that labels obtained from shape-based clustering improved the prediction more than feature-based cluster labels. The prediction errors were minimized with the proposed DNN model, where static features are introduced to the first and second layers twice.

Self-Reported Late Effects in Childhood Cancer Survivors in Kenya

Background: The number of children surviving cancer in low and middle-income countries is expected to grow in the coming years. Knowledge about late effects and follow-up preferences in Kenya is lacking.Objectives: This study assessed self-reported late effects in Kenyan childhood cancer survivors and explored their preferences for survivorship care.Methods: Childhood cancer survivors, having successfully completed treatment for at least one year, were interviewed using semi-structured questionnaires during clinic or home visits between 2021-2022. Medical records were reviewed for patient and treatment characteristics.Results: Twenty-six survivors of hematological malignancies (n=19, 73%), solid tumors (n=6, 23%), unknown tumor type (n=1, 4%), were interviewed. Most survivors (n=19, 73%) solely received chemotherapy and one survivor (4%) was irradiated. Median time since treatment completion was seven years. Fifteen survivors (58%) were previously lost to follow-up. Many survivors (n=19; 73%) self-reported late effects, predominantly pain and fatigue. Survivors (n=11, 42%) were limited in daily life activities: physical work (n=10, 38%), personal care (n=6, 23%), social activities (n=6, 23%). Eight survivors (31%) recalled being informed about late effects. Some survivors experienced a negative attitude toward cancer in regional hospitals. Follow-up duration was longer among informed patients (p=0.043). Survivors recommended education and survivor meetings and preferred their follow-up to be done at the referral center.Conclusions: Kenyan childhood cancer survivors self-report late effects, comparable in frequency, nature and severity to other survivors worldwide. Survivors and healthcare providers require education about the lifelong impact of childhood cancer and should have access to survivorship expertise to continue follow-up. Keywords: Pediatrics; survivorship; neoplasms; aftercare.

A Longitudinal Study of Escherichia coli Clinical Isolates from the Tracheal Aspirates of a Paediatric Patient-Strain Type Similar to Pandemic ST131.

Escherichia coli is a Gram-negative bacterium and part of the intestinal microbiota. However, it can cause various diarrhoeal illnesses, i.e., traveller's diarrhoea, dysentery, and extraintestinal infections when the bacteria are translocated from the intestine to other organs, such as urinary tract infections, abdominal and pelvic infections, pneumonia, bacteraemia, and meningitis. It is also an important pathogen in intensive care units where cross-infection may cause intrahospital spread with serious consequences. Within this study, four E. coli isolates from the tracheal aspirates of a tracheotomised paediatric patient on chronic respiratory support were analysed and compared for antibiotic resistance and virulence potential. Genomes of all four isolates (5381a, 5381b, 5681, 5848) were sequenced using Oxford Nanopore Technology. According to PFGE analysis, the clones of isolates 5681 and 5848 were highly similar, and differ from 5381a and 5381b which were isolated first chronologically. All four E. coli isolates belonged to an unknown sequence type, related to the E. coli ST131, a pandemic clone that is evolving rapidly with increasing levels of antimicrobial resistance. All four E. coli isolates in this study exhibited a multidrug-resistant phenotype as, according to MIC data, they were resistant to ceftriaxone, ciprofloxacin, doxycycline, minocycline, and tetracycline. In addition, principal component analyses revealed that isolates 5681 and 5848, which were recovered later than 5381a and 5381b (two weeks and three weeks, respectively) possessed more complex antibiotic resistance genes and virulence profiles, which is concerning considering the short time period during which the strains were isolated.

Carboxy-terminal polyglutamylation regulates signaling and phase separation of the Dishevelled protein

Polyglutamylation is a reversible posttranslational modification that is catalyzed by enzymes of the tubulin tyrosine ligase-like (TTLL) family. Here, we found that TTLL11 generates a previously unknown type of polyglutamylation that is initiated by the addition of a glutamate residue to the free C-terminal carboxyl group of a substrate protein. TTLL11 efficiently polyglutamylates the Wnt signaling protein Dishevelled 3 (DVL3), thereby changing the interactome of DVL3. Polyglutamylation increases the capacity of DVL3 to get phosphorylated, to undergo phase separation, and to act in the noncanonical Wnt pathway. Both carboxy-terminal polyglutamylation and the resulting reduction in phase separation capacity of DVL3 can be reverted by the deglutamylating enzyme CCP6, demonstrating a causal relationship between TTLL11-mediated polyglutamylation and phase separation. Thus, C-terminal polyglutamylation represents a new type of posttranslational modification, broadening the range of proteins that can be modified by polyglutamylation and providing the first evidence that polyglutamylation can modulate protein phase separation.

A Vehicle Abnormal Behavior Detection Model in Single Intelligent Vehicle Scenarios

Connected and automated vehicles (CAVs) play a vital role in transforming human mobility, tackling road congestion and road safety. However, CAVs rely heavily on the security, accuracy, and stability of sensor readings and network data. When there are anomalies in the data, it is necessary to detect them in a timely manner and handle them. However, under single intelligent vehicle scenarios, existing detection methods often struggle to identify unknown types of anomalies and are difficult to deploy on computationally limited vehicle terminals. To address the aforementioned issues, this paper proposes a vehicle anomaly data detection method based on deep learning. First, we modify the discriminator based on the GAN network, so that the network can assign different weights to different sensors, thus improving the generalization performance of the model. Afterwards, we assign weights to each parameter of the model during the training process, and then prune the model according to the weights to improve its computational speed. We verify the reliability of our method on the Safe Pilot Model Deployment (SPMD) data set. It is shown that the proposed model has good detection performance for various anomaly data, especially when facing data that were not encountered during the training process, and the proposed model effectively reduces the computational time of the detection process.

The Adaptive Physical Activity Study for Stroke: A Feasibility Sequential Multiple Assignment Randomized Trial

Abstract Background Between 1990 and 2017, stroke cases nearly doubled, ranking as the second leading cause of death and disability globally. Inadequate physical activity (PA) is a leading risk factor. Recent guidelines advise stroke survivors to follow population-based PA guidelines. Given stroke heterogeneity personalised PA interventions are recommended. Therefore, this study used a Sequential Multiple-Assignment Randomized Trial (SMART) design to develop an adaptive mHealth intervention to improve PA post-stroke. Methods This SMART is in progress, with recruitment completed and follow-up data ending November 2024. A 12-week intervention programme was developed incorporating public patient involvement (PPI), empirical evidence and behavioural change theory to increase participation in Structured Exercise (SE), Lifestyle Physical Activity (LPA), or both. The study enrolled 54 ambulatory, community-dwelling people who were randomly assigned to either SE or LPA treatment groups. Participants used Fitbit Inspire 2 devices to monitor step counts and set personalized weekly step goals. At six weeks post-randomisation, participants were evaluated and classified as responders or non-responders based on their step count target achievement. Non-responders were re-randomized to switch treatments or augment their current regime with an additional intervention. Responders maintained their original treatment protocol. Primary outcomes address recruitment, retention and adherence rates. Secondary outcomes encompass overall health. Results Preliminary results include 54 participants (56% male) aged 51.58 (± 15.12) with Ischaemic (44%), Haemorrhagic (28%) or unknown (28%) stroke type at baseline. The mean weekly step was 7046 (±3797) while the mean Fatigue Severity Scale score was 38.02 (± 14.60). Participants had a mean Hospital Anxiety and Depression Scale score of 12.33 (± 6.98). Additionally, the mean Stroke Specific Quality of Life was 155.04 (58.84). Conclusion This is the first trial to use a SMART to design an adaptive intervention post-stroke internationally. Findings will inform the development of a PA intervention and subsequently will be evaluated against treatment as usual in a definitive trial.

Multi-view learning framework for predicting unknown types of cancer markers via directed graph neural networks fitting regulatory networks.

The discovery of diagnostic and therapeutic biomarkers for complex diseases, especially cancer, has always been a central and long-term challenge in molecular association prediction research, offering promising avenues for advancing the understanding of complex diseases. To this end, researchers have developed various network-based prediction techniques targeting specific molecular associations. However, limitations imposed by reductionism and network representation learning have led existing studies to narrowly focus on high prediction efficiency within single association type, thereby glossing over the discovery of unknown types of associations. Additionally, effectively utilizing network structure to fit the interaction properties of regulatory networks and combining specific case biomarker validations remains an unresolved issue in cancer biomarker prediction methods. To overcome these limitations, we propose a multi-view learning framework, CeRVE, based on directed graph neural networks (DGNN) for predicting unknown type cancer biomarkers. CeRVE effectively extracts and integrates subgraph information through multi-view feature learning. Subsequently, CeRVE utilizes DGNN to simulate the entire regulatory network, propagating node attribute features and extracting various interaction relationships between molecules. Furthermore, CeRVE constructed a comparative analysis matrix of three cancers and adjacent normal tissues through The Cancer Genome Atlas and identified multiple types of potential cancer biomarkers through differential expression analysis of mRNA, microRNA, and long noncoding RNA. Computational testing of multiple types of biomarkers for 72 cancers demonstrates that CeRVE exhibits superior performance in cancer biomarker prediction, providing a powerful tool and insightful approach for AI-assisted disease biomarker discovery.

A multichannel graph neural network based on multisimilarity modality hypergraph contrastive learning for predicting unknown types of cancer biomarkers.

Identifying potential cancer biomarkers is a key task in biomedical research, providing a promising avenue for the diagnosis and treatment of human tumors and cancers. In recent years, several machine learning-based RNA-disease association prediction techniques have emerged. However, they primarily focus on modeling relationships of a single type, overlooking the importance of gaining insights into molecular behaviors from a complete regulatory network perspective and discovering biomarkers of unknown types. Furthermore, effectively handling local and global topological structural information of nodes in biological molecular regulatory graphs remains a challenge to improving biomarker prediction performance. To address these limitations, we propose a multichannel graph neural network based on multisimilarity modality hypergraph contrastive learning (MML-MGNN) for predicting unknown types of cancer biomarkers. MML-MGNN leverages multisimilarity modality hypergraph contrastive learning to delve into local associations in the regulatory network, learning diverse insights into the topological structures of multiple types of similarities, and then globally modeling the multisimilarity modalities through a multichannel graph autoencoder. By combining representations obtained from local-level associations and global-level regulatory graphs, MML-MGNN can acquire molecular feature descriptors benefiting from multitype association properties and the complete regulatory network. Experimental results on predicting three different types of cancer biomarkers demonstrate the outstanding performance of MML-MGNN. Furthermore, a case study on gastric cancer underscores the outstanding ability of MML-MGNN to gain deeper insights into molecular mechanisms in regulatory networks and prominent potential in cancer biomarker prediction.

A multi-task prediction method based on neighborhood structure embedding and signed graph representation learning to infer the relationship between circRNA, miRNA, and cancer.

Research shows that competing endogenous RNA is widely involved in gene regulation in cells, and identifying the association between circular RNA (circRNA), microRNA (miRNA), and cancer can provide new hope for disease diagnosis, treatment, and prognosis. However, affected by reductionism, previous studies regarded the prediction of circRNA-miRNA interaction, circRNA-cancer association, and miRNA-cancer association as separate studies. Currently, few models are capable of simultaneously predicting these three associations. Inspired by holism, we propose a multi-task prediction method based on neighborhood structure embedding and signed graph representation learning, CMCSG, to infer the relationship between circRNA, miRNA, and cancer. Our method aims to extract feature descriptors of all molecules from the circRNA-miRNA-cancer regulatory network using known types of association information to predict unknown types of molecular associations. Specifically, we first constructed the circRNA-miRNA-cancer association network (CMCN), which is constructed based on the experimentally verified biomedical entity regulatory network; next, we combine topological structure embedding methods to extract feature representations in CMCN from local and global perspectives, and use denoising autoencoder for enhancement; then, combined with balance theory and state theory, molecular features are extracted from the point of social relations through the propagation and aggregation of signed graph attention network; finally, the GBDT classifier is used to predict the association of molecules. The results show that CMCSG can effectively predict the relationship between circRNA, miRNA, and cancer. Additionally, the case studies also demonstrate that CMCSG is capable of accurately identifying biomarkers across various types of cancer. The data and source code can be found at https://github.com/1axin/CMCSG.