Robust Diagnostic Model Research Articles

Colon Cancer Disease Diagnosis Based on Convolutional Neural Network and Fishier Mantis Optimizer.

Colon cancer is a prevalent and potentially fatal disease that demands early and accurate diagnosis for effective treatment. Traditional diagnostic approaches for colon cancer often face limitations in accuracy and efficiency, leading to challenges in early detection and treatment. In response to these challenges, this paper introduces an innovative method that leverages artificial intelligence, specifically convolutional neural network (CNN) and Fishier Mantis Optimizer, for the automated detection of colon cancer. The utilization of deep learning techniques, specifically CNN, enables the extraction of intricate features from medical imaging data, providing a robust and efficient diagnostic model. Additionally, the Fishier Mantis Optimizer, a bio-inspired optimization algorithm inspired by the hunting behavior of the mantis shrimp, is employed to fine-tune the parameters of the CNN, enhancing its convergence speed and performance. This hybrid approach aims to address the limitations of traditional diagnostic methods by leveraging the strengths of both deep learning and nature-inspired optimization to enhance the accuracy and effectiveness of colon cancer diagnosis. The proposed method was evaluated on a comprehensive dataset comprising colon cancer images, and the results demonstrate its superiority over traditional diagnostic approaches. The CNN-Fishier Mantis Optimizer model exhibited high sensitivity, specificity, and overall accuracy in distinguishing between cancer and non-cancer colon tissues. The integration of bio-inspired optimization algorithms with deep learning techniques not only contributes to the advancement of computer-aided diagnostic tools for colon cancer but also holds promise for enhancing the early detection and diagnosis of this disease, thereby facilitating timely intervention and improved patient prognosis. Various CNN designs, such as GoogLeNet and ResNet-50, were employed to capture features associated with colon diseases. However, inaccuracies were introduced in both feature extraction and data classification due to the abundance of features. To address this issue, feature reduction techniques were implemented using Fishier Mantis Optimizer algorithms, outperforming alternative methods such as Genetic Algorithms and simulated annealing. Encouraging results were obtained in the evaluation of diverse metrics, including sensitivity, specificity, accuracy, and F1-Score, which were found to be 94.87%, 96.19%, 97.65%, and 96.76%, respectively.

Identification of schizophrenia by applying interpretable radiomics modeling with structural magnetic resonance imaging of the cerebellum.

The cerebellum is involved in higher-order mental processing as well as sensorimotor functions. Although structural abnormalities in the cerebellum have been demonstrated in schizophrenia, neuroimaging techniques are not yet applicable to identify them given the lack of biomarkers. We aimed to develop a robust diagnostic model for schizophrenia using radiomic features from T1-weighted magnetic resonance imaging (T1-MRI) of the cerebellum. A total of 336 participants (174 schizophrenia; 162 healthy controls [HCs]) were allocated to training (122 schizophrenia; 115 HCs) and test (52 schizophrenia; 47 HCs) cohorts. We obtained 2568 radiomic features from T1-MRI of the cerebellar subregions. After feature selection, a light gradient boosting machine classifier was trained. The discrimination and calibration of the model were evaluated. SHapley Additive exPlanations (SHAP) was applied to determine model interpretability. We identified 17 radiomic features to differentiate participants with schizophrenia from HCs. In the test cohort, the radiomics model had an area under the curve, accuracy, sensitivity, and specificity of 0.89 (95% confidence interval: 0.82-0.95), 78.8%, 88.5%, and 75.4%, respectively. The model explanation by SHAP suggested that the second-order size zone non-uniformity feature from the right lobule IX and first-order energy feature from the right lobules V and VI were highly associated with the risk of schizophrenia. The radiomics model focused on the cerebellum demonstrates robustness in diagnosing schizophrenia. Our results suggest that microcircuit disruption in the posterior cerebellum is a disease-defining feature of schizophrenia, and radiomics modeling has potential for supporting biomarker-based decision-making in clinical practice.

Exploration and verification a 13-gene diagnostic framework for ulcerative colitis across multiple platforms via machine learning algorithms

Ulcerative colitis (UC) is a chronic inflammatory bowel disease with intricate pathogenesis and varied presentation. Accurate diagnostic tools are imperative to detect and manage UC. This study sought to construct a robust diagnostic model using gene expression profiles and to identify key genes that differentiate UC patients from healthy controls. Gene expression profiles from eight cohorts, encompassing a total of 335 UC patients and 129 healthy controls, were analyzed. A total of 7530 gene sets were computed using the GSEA method. Subsequent batch correction, PCA plots, and intersection analysis identified crucial pathways and genes. Machine learning, incorporating 101 algorithm combinations, was employed to develop diagnostic models. Verification was done using four external cohorts, adding depth to the sample repertoire. Evaluation of immune cell infiltration was undertaken through single-sample GSEA. All statistical analyses were conducted using R (Version: 4.2.2), with significance set at a P value below 0.05. Employing the GSEA method, 7530 gene sets were computed. From this, 19 intersecting pathways were discerned to be consistently upregulated across all cohorts, which pertained to cell adhesion, development, metabolism, immune response, and protein regulation. This corresponded to 83 unique genes. Machine learning insights culminated in the LASSO regression model, which outperformed others with an average AUC of 0.942. This model's efficacy was further ratified across four external cohorts, with AUC values ranging from 0.694 to 0.873 and significant Kappa statistics indicating its predictive accuracy. The LASSO logistic regression model highlighted 13 genes, with LCN2, ASS1, and IRAK3 emerging as pivotal. Notably, LCN2 showcased significantly heightened expression in active UC patients compared to both non-active patients and healthy controls (P < 0.05). Investigations into the correlation between these genes and immune cell infiltration in UC highlighted activated dendritic cells, with statistically significant positive correlations noted for LCN2 and IRAK3 across multiple datasets. Through comprehensive gene expression analysis and machine learning, a potent LASSO-based diagnostic model for UC was developed. Genes such as LCN2, ASS1, and IRAK3 hold potential as both diagnostic markers and therapeutic targets, offering a promising direction for future UC research and clinical application.

Culture of Cultures: A Small Step Towards Augmenting Diagnostic Stewardship.

Laboratories with a well-established diagnostic stewardship program for culture and the antimicrobial susceptibility test (C-AST) play a key role in guiding clinicians to institute specific targeted therapy. Blood culture is one of the most critical investigations performed at a microbiology laboratory. Therefore, it is particularly important to develop a robust diagnostic stewardship model for the blood culture laboratory division. Aiming at this hypothesis, this hospital-based clinical audit carried out within the Department of General Medicine, centered on the critical domain of antimicrobial stewardship (AMSP). The audit's primary objective was to systematically evaluate prevailing practices and pinpoint areas necessitating refinement in the administration of antimicrobial agents. Employing a meticulous approach involving exhaustive data scrutiny and feedback mechanisms, the audit unearthed strategic opportunities to optimize prescription patterns, curtail unwarranted antimicrobial utilization, and fortify adherence to established guidelines. The subsequent execution of targeted interventions, encompassing educational initiatives and routine performance feedback, culminated in a noteworthy enhancement of antimicrobial prescribing practices. These outcomes unequivocally underscore the efficacy of the audit in cultivating a milieu of judicious antimicrobial utilization, thereby augmenting patient care and mitigating antibiotic resistance within the department.

Machine learning and deep learning to identifying subarachnoid haemorrhage macrophage-associated biomarkers by bulk and single-cell sequencing.

We investigated subarachnoid haemorrhage (SAH) macrophage subpopulations and identified relevant key genes for improving diagnostic and therapeutic strategies. SAH rat models were established, and brain tissue samples underwent single-cell transcriptome sequencing and bulk RNA-seq. Using single-cell data, distinct macrophage subpopulations, including a unique SAH subset, were identified. The hdWGCNA method revealed 160 key macrophage-related genes. Univariate analysis and lasso regression selected 10 genes for constructing a diagnostic model. Machine learning algorithms facilitated model development. Cellular infiltration was assessed using the MCPcounter algorithm, and a heatmap integrated cell abundance and gene expression. A 3 × 3 convolutional neural network created an additional diagnostic model, while molecular docking identified potential drugs. The diagnostic model based on the 10 selected genes achieved excellent performance, with an AUC of 1 in both training and validation datasets. The heatmap, combining cell abundance and gene expression, provided insights into SAH cellular composition. The convolutional neural network model exhibited a sensitivity and specificity of 1 in both datasets. Additionally, CD14, GPNMB, SPP1 and PRDX5 were specifically expressed in SAH-associated macrophages, highlighting its potential as a therapeutic target. Network pharmacology analysis identified some targeting drugs for SAH treatment. Our study characterised SAH macrophage subpopulations and identified key associated genes. We developed a robust diagnostic model and recognised CD14, GPNMB, SPP1 and PRDX5 as potential therapeutic targets. Further experiments and clinical investigations are needed to validate these findings and explore the clinical implications of targets in SAH treatment.

A Machine Learning Model for Predicting Major Depressive Disorder Using Diffusion-Tensor Imaging Data

IntroductionMajor Depressive Disorder (MDD) stands as a prevalent psychiatric condition within the general population. Despite extensive research efforts, the identification of definitive diagnostic biomarkers for depressive disorders remains elusive. Currently, machine learning methods are gaining prominence in the diagnosis of medical illnesses.ObjectivesThis study aims to construct a machine learning-based prediction model for Major Depressive Disorder (MDD) by harnessing diffusion tensor imaging (DTI) data.MethodsThe DTI datasets comprising MDD (N=83) and Healthy Control (N=70) groups were procured from the cohort study of Anxiety and Depression conducted at the National Center for Mental Health in South Korea. A machine learning method using a decision tree algorithm was employed to select relevant brain regions and establish a robust diagnostic model. Features associated with white matter (WM) tracts were chosen through recursive feature elimination.Results Demographic characteristics, including age, sex, and handedness, displayed no significant differences between the MDD and Healthy Control groups. However, the total score of the Beck Depression Inventory was notably higher in individuals with MDD compared to Healthy Controls. A diagnostic model was crafted using the decision tree algorithms to distinguish between the two groups. The model demonstrated the following classification performance metrics: accuracy (65.6% ± 8.5), sensitivity (66.6% ± 12.5), and specificity (64.7% ± 13.6). Furthermore, through recursive feature elimination, specific neuroanatomical features tied to brain structures such as the inferior cerebellar peduncle, posterior thalamic radiation, cingulum (hippocampus), uncinate fasciculus, and tapetum were identified.ConclusionsDespite of limited performance of classification, a machine learning-based approach could provide insights into the development of a diagnostic model for MDD using neuroimaging data. Furthermore, these features, derived from DTI-derived data, may have implications for understanding the neural underpinnings of major depressive disorder.Disclosure of InterestNone Declared

Integrated analysis of m6A regulator-mediated RNA methylation modification patterns and immune characteristics in Sjögren's syndrome

The epigenetic modifier N6-methyladenosine (m6A), recognized as the most prevalent internal modification in messenger RNA (mRNA), has recently emerged as a pivotal player in immune regulation. Its dysregulation has been implicated in the pathogenesis of various autoimmune conditions. However, the implications of m6A modification within the immune microenvironment of Sjögren's syndrome (SS), a chronic autoimmune disorder characterized by exocrine gland dysfunction, remain unexplored. Herein, we leverage an integrative analysis combining public database resources and novel sequencing data to investigate the expression profiles of m6A regulatory genes in SS. Our cohort comprised 220 patients diagnosed with SS and 62 healthy individuals, enabling a comprehensive evaluation of peripheral blood at the transcriptomic level. We report a significant association between SS and altered expression of key m6A regulators, with these changes closely tied to the activation of CD4+ T cells. Employing a random forest (RF) algorithm, we identified crucial genes contributing to the disease phenotype, which facilitated the development of a robust diagnostic model via multivariate logistic regression analysis. Further, unsupervised clustering revealed two distinct m6A modification patterns, which were significantly associated with variations in immunocyte infiltration, immune response activity, and biological function enrichment in SS. Subsequently, we proceeded with a screening process aimed at identifying genes that were differentially expressed (DEGs) between the two groups distinguished by m6A modification. Leveraging these DEGs, we employed weight gene co-expression network analysis (WGCNA) to uncover sets of genes that exhibited strong co-variance and hub genes that were closely linked to m6A modification. Through rigorous analysis, we identified three critical m6A regulators - METTL3, ALKBH5, and YTHDF1 - alongside two m6A-related hub genes, COMMD8 and SRP9. These elements collectively underscore a complex but discernible pattern of m6A modification that appears to be integrally linked with SS's pathogenesis. Our findings not only illuminate the significant correlation between m6A modification and the immune microenvironment in SS but also lay the groundwork for a deeper understanding of m6A regulatory mechanisms. More importantly, the identification of these key regulators and hub genes opens new avenues for the diagnosis and treatment of SS, presenting potential targets for therapeutic intervention.

Exploring m6A-linked aging genes in osteoarthritis and broad cancer spectrum: Prospects for diagnostic and therapeutic advancements.

Osteoarthritis (OA) is a prevalent degenerative joint disease that significantly impacts individuals and healthcare systems worldwide. However, the exploration of N6-methyladenosine (m6A)-related aging genes in OA pathogenesis remains largely underexplored. This study aimed to elucidate the role of m6A-related aging genes in OA and to develop a robust diagnostic model based on their expression profiles. Leveraging publicly available gene expression datasets, we conducted consensus clustering to categorize OA into distinct subtypes, guided by the expression patterns of m6A-related aging genes. Utilizing XGBoost, a cutting-edge machine learning approach, we identified key diagnostic genes and constructed a predictive model. Our investigation extended to the immune functions of these genes, shedding light on potential therapeutic targets and underlying regulatory mechanisms. Our analysis unveiled specific OA subtypes, each marked by unique expression profiles of m6A-related aging genes. We pinpointed a set of pivotal diagnostic genes, offering potential therapeutic avenues. The developed diagnostic model exhibited exceptional capability in distinguishing OA patients from healthy controls. To corroborate our computational findings, we performed quantitative real-time polymerase chain reaction analyses on two cell lines: HC-OA (representing adult osteoarthritis cells) and C-28/I2 (representative of normal human chondrocytes). The gene expression patterns observed were consistent with our bioinformatics predictions, further validating our initial results. In conclusion, this study underscores the significance of m6A-related aging genes as promising biomarkers for diagnosis and prognosis, as well as potential therapeutic targets in OA. Although these findings are encouraging, further validation and functional analyses are crucial for their clinical application.

Identification of Co-diagnostic Genes for Heart Failure and Hepatocellular Carcinoma Through WGCNA and Machine Learning Algorithms.

This research delves into the intricate relationship between hepatocellular carcinoma (HCC) and heart failure (HF) by exploring shared genetic characteristics and molecular processes. Employing advanced methodologies such as differential analysis, weighted correlation network analysis (WGCNA), and algorithms like Random Forest (RF), Least Absolute Shrinkage Selection (LASSO), and XGBoost, we meticulously identified modular differential genes (DEGs) associated with both HF and HCC. Gene Set Variation Analysis (GSVA) and single sample gene set enrichment analysis (ssGSEA) were employed to unveil underlying biological mechanisms. The study revealed 88 core genes shared between HF and HCC, indicating a common mechanism. Enrichment analysis emphasized the roles of immune responses and inflammation in both diseases. Leveraging XGBoost, we crafted a robust multigene diagnostic model (including FCN3, MAP2K1, AP3M2, CDH19) with an area under the curve (AUC)>0.9, showcasing exceptional predictive accuracy. GSVA and ssGSEA analyses unveiled the involvement of immune cells and metabolic pathways in the pathogenesis of HF and HCC. This research uncovers a pivotal interplay between HF and HCC, highlighting shared pathways and key genes, offering promising insights for future clinical treatments and experimental research endeavors.

Identification and validation of platelet-related diagnostic markers and potential drug screening in ischemic stroke by integrating comprehensive bioinformatics analysis and machine learning.

Ischemic stroke (IS), caused by blood and oxygen deprivation due to cerebral thrombosis, has links to activated and aggregated platelets. Discovering platelet-related biomarkers, developing diagnostic models, and screening antiplatelet drugs are crucial for IS diagnosis and treatment. Combining and normalizing GSE16561 and GSE22255 datasets identified 1,753 upregulated and 1,187 downregulated genes. Fifty-one genes in the platelet-related module were isolated using weighted gene co-expression network analysis (WGCNA) and other analyses, including 50 upregulated and one downregulated gene. Subsequent enrichment and network analyses resulted in 25 platelet-associated genes and six diagnostic markers for a risk assessment model. This model's area under the ROC curve outperformed single genes, and in the peripheral blood of the high-risk group, immune infiltration indicated a higher proportion of CD4, resting CD4 memory, and activated CD4 memory T cells, along with a lower proportion of CD8 T cells in comparison to the low-risk group. Utilizing the gene expression matrix and the CMap database, we identified two potential drugs for IS. Finally, a rat MACO/R model was used to validate the diagnostic markers' expression and the drugs' predicted anticoagulant effects. We identified six IS platelet-related biomarkers (APP, THBS1, F13A1, SRC, PPBP, and VCL) for a robust diagnostic model. The drugs alpha-linolenic acid and ciprofibrate have potential antiplatelet effects in IS. This study advances early IS diagnosis and treatment.

A comprehensive analysis of genes associated with hypoxia and cuproptosis in pulmonary arterial hypertension using machine learning methods and immune infiltration analysis: AHR is a key gene in the cuproptosis process.

Pulmonary arterial hypertension (PAH) is a serious condition characterized by elevated pulmonary artery pressure, leading to right heart failure and increased mortality. This study investigates the link between PAH and genes associated with hypoxia and cuproptosis. We utilized expression profiles and single-cell RNA-seq data of PAH from the GEO database and genecad. Genes related to cuproptosis and hypoxia were identified. After normalizing the data, differential gene expression was analyzed between PAH and control groups. We performed clustering analyses on cuproptosis-related genes and constructed a weighted gene co-expression network (WGCNA) to identify key genes linked to cuproptosis subtype scores. KEGG, GO, and DO enrichment analyses were conducted for hypoxia-related genes, and a protein-protein interaction (PPI) network was created using STRING. Immune cell composition differences were examined between groups. SingleR and Seurat were used for scRNA-seq data analysis, with PCA and t-SNE for dimensionality reduction. We analyzed hub gene expression across single-cell clusters and built a diagnostic model using LASSO and random forest, optimizing parameters with 10-fold cross-validation. A total of 113 combinations of 12 machine learning algorithms were employed to evaluate model accuracy. GSEA was utilized for pathway enrichment analysis of AHR and FAS, and a Nomogram was created to assess risk impact. We also analyzed the correlation between key genes and immune cell types using Spearman correlation. We identified several diagnostic genes for PAH linked to hypoxia and cuproptosis. PPI networks illustrated relationships among these hub genes, with immune infiltration analysis highlighting associations with monocytes, macrophages, and CD8 T cells. The genes AHR, FAS, and FGF2 emerged as key markers, forming a robust diagnostic model (NaiveBayes) with an AUC of 0.9. AHR, FAS, and FGF2 were identified as potential biomarkers for PAH, influencing cell proliferation and inflammatory responses, thereby offering new insights for PAH prevention and treatment.

A diagnostic prediction model for the early detection of heart failure following primary percutaneous coronary intervention in patients with ST-elevation myocardial infarction.

In this study, we aimed to construct a robust diagnostic model that can predict the early onset of heart failure in patients with ST-elevation myocardial infarction (STEMI) following a primary percutaneous coronary intervention (PCI). This diagnostic model can facilitate the early stratification of high-risk patients, thereby optimizing therapeutic management. We performed a retrospective analysis of 664 patients with STEMI who underwent their inaugural PCI. We performed logistic regression along with optimal subset regression and identified important risk factors associated with the early onset of heart failure during the time of admission. Based on these determinants, we constructed a predictive model and confirmed its diagnostic precision using a receiver operating characteristic (ROC) curve. The logistic and optimal subset regression analyses revealed the following three salient risk factors crucial for the early onset of heart failure: the Killip classification, the presence of renal insufficiency, and increased troponin T levels. The constructed prognostic model exhibited excellent discriminative ability, which was indicated by an area under the curve value of 0.847. The model's 95% confidence interval following 200 Bootstrap iterations was found to be between 0.767 and 0.925. The Hosmer-Lemeshow test revealed a chi-square value of 3.553 and a p-value of 0.938. Notably, the calibration of the model remained stable even after 500 Bootstrap evaluations. Furthermore, decision curve analysis revealed a substantial net benefit of the model. We have successfully constructed a diagnostic prediction model to predict the incipient stages of heart failure in patients with STEMI following primary PCI. This diagnostic model can revolutionize patient care, allowing clinicians to quickly identify and create individualized interventions for patients at a higher risk.

A comprehensive analysis of immune features and construction of an immune gene diagnostic model for sepsis

Sepsis is a life-threatening syndrome resulting from immune system dysfunction that is caused by infection. It is of great importance to analyze the immune characteristics of sepsis, identify the key immune system related genes, and construct diagnostic models for sepsis. In this study, the sepsis transcriptome and expression profiling data were merged into an integrated dataset containing 277 sepsis samples and 117 non-sepsis control samples. Single-sample gene set enrichment analysis (ssGSEA) was used to assess the immune cell infiltration. Two sepsis immune subtypes were identified based on the 22 differential immune cells between the sepsis and the healthy control groups. Weighted gene co-expression network analysis (WCGNA) was used to identify the key module genes. Then, 36 differentially expressed immune-related genes were identified, based on which a robust diagnostic model was constructed with 11 diagnostic genes. The expression of 11 diagnostic genes was finally assessed in the training and validation datasets respectively. In this study, we provide comprehensive insight into the immune features of sepsis and establish a robust diagnostic model for sepsis. These findings may provide new strategies for the early diagnosis of sepsis in the future.

Gene methylation status in focus of advanced prostate cancer diagnostics and improved individual outcomes.

Prostate cancer (PCa) is the most prevalent type of male genitourinary tumor, remains the second leading cause of deaths due to cancer in the United States in men. The aim of this study was to perform an integrative epigenetic analysis to explore the epigenetic abnormalities involved in the development and progression of PCa, and present advanced diagnostics and improved individual outcomes. Genome-wide DNA methylation profiles obtained from The Cancer Genome Atlas (TCGA) were analyzed and a diagnostic model was constructed. For validation, we employed profiles from the Gene Expression Omnibus (GEO) and methylation data derived from clinical samples. Gene set enrichment analysis (GSEA) and the Tumor Immune Estimation Resource (TIMER) were employed for GSEA and to assess immune cell infiltration, respectively. An accurate diagnostic method for PCa was established based on the methylation level of Cyclin-D2 (CCND2) and glutathione S-transferase pi-1 (GSTP1), with an impressive area under the curve (AUC) value of 0.937. The model's reliability was further confirmed through validation using four GEO datasets GSE76938 (AUC =0.930), GSE26126 (AUC =0.906), GSE112047 (AUC =1.000), GSE84749 (AUC =0.938) and clinical samples (AUC =0.980). Notably, the TIMER analysis indicated that hypermethylation of CCND2 and GSTP1 was associated with reduced immune cell infiltration, higher tumor purity, and an increased risk of tumor progression. In conclusion, our study provides a robust and reliable methylation-based diagnostic model for PCa. This model holds promise as an improved approach for screening and diagnosing PCa, potentially enhancing early detection and patient outcomes, as well as for an advanced clinical management for PCa in the framework of predictive, preventive and personalised medicine.

Integrating fluid–solid coupling domain knowledge with deep learning models: An automatic and interpretable diagnostic system for the silting disease of drainage pipelines

An accurate and robust diagnostic model for the silting disease of drainage pipelines is significant for the numerical simulation of urban waterlogging and assessment of risk areas. Various long short-term memory (LSTM) neural networks have been adopted for intelligently diagnosing pipeline siltation owing to the strong correlation between pipeline siltation and flow characteristics over time. However, LSTM often has strong randomness and uncertainty when diagnosing the collocation points of non-training set features and rules and violates the basic fluid–solid coupling physical mechanism. To fully integrate fluid–solid coupling domain knowledge with data-driven neural networks to improve the accuracy of the diagnostic model, this study proposes a knowledge-data collaboratively driven model based on hard constraint projection (HCP) and theory-guided loss function correction (TLFC) for the intelligent diagnosis of drainage pipeline siltation (IDPS). This diagnostic model converts physical fluid–solid coupling constraints, such as the governing equations and boundary/initial conditions, into a mathematical simplification that is easy to handle through discretisation. It adjusts input data sequence and optimises diagnostic results from intelligent algorithms through HCP on the hyperplane and theory-guided loss function correction. The performance of the self-diagnosing system integrating fluid–solid coupling domain knowledge with deep learning models was verified through experiments based on a full-scale prototype. The experiment results indicate that the proposed model had a good ability of robustness to resist the noisy input observations when no less than 600 boundary points were adopted. Compared with typical LSTM, TLFC-based LSTM and HCP-based LSTM, the proposed algorithm achieved highest performance regarding diagnostic accuracy. The mean absolute percentage error of the proposed system was kept below 2%. Furthermore, owing to the knowledge-data collaboratively driven mechanism, the proposed system can extrapolate and accurately diagnose situations outside the training dataset range.

Data-driven prognosis of failure detection and prediction of lithium-ion batteries

Battery prognostics and health management predictive models are essential components of safety and reliability protocols in battery management system frameworks. Overall, developing a robust and efficient fault diagnostic battery model that aligns with the current literature is an essential step in ensuring the safety of battery function. For this purpose, a multi-physics, multi-scale deterministic data-driven prognosis (DDP) is proposed that only relies on in situ measurements of data and estimates the failure based on the curvature information extracted from the system. Unlike traditional applications that require explicit expression of conservation principle to represent the system's behavior, the proposed method devices a local conservation functional in the neighborhood of each data point which is represented as the minimization of curvature in the system. Pursuing such a deterministic approach, DDP eliminates the need for offline training regimen by considering only two consecutive time instances to make the prognostication that are sufficient to extract the behavioral pattern of the system. The developed framework is then employed to analyze the health of lithium ion batteries by monitoring the performance and detecting faults within the system's behavior. Based on the outcomes, the DDP exhibits promising results in detection of anomaly and prognostication of batteries' failure.

Hagnifinder: Recovering magnification information of digital histological images using deep learning

Background and objectiveTraining a robust cancer diagnostic or prognostic artificial intelligent model using histology images requires a large number of representative cases with labels or annotations, which are difficult to obtain. The histology snapshots available in published papers or case reports can be used to enrich the training dataset. However, the magnifications of these invaluable snapshots are generally unknown, which limits their usage. Therefore, a robust magnification predictor is required for utilizing those diverse snapshot repositories consisting of different diseases. This paper presents a magnification prediction model named Hagnifinder for H&E-stained histological images. MethodsHagnifinder is a regression model based on a modified convolutional neural network (CNN) that contains 3 modules: Feature Extraction Module, Regression Module, and Adaptive Scaling Module (ASM). In the training phase, the Feature Extraction Module first extracts the image features. Secondly, the ASM is proposed to address the learned feature values uneven distribution problem. Finally, the Regression Module estimates the mapping between the regularized extracted features and the magnifications. We construct a new dataset for training a robust model, named Hagni40, consisting of 94 643 H&E-stained histology image patches at 40 different magnifications of 13 types of cancer based on The Cancer Genome Atlas. To verify the performance of the Hagnifinder, we measure the accuracy of the predictions by setting the maximum allowable difference values (0.5, 1, and 5) between the predicted magnification and the actual magnification. We compare Hagnifinder with state-of-the-art methods on a public dataset BreakHis and the Hagni40. ResultsThe Hagnifinder provides consistent prediction accuracy, with a mean accuracy of 98.9%, across 40 different magnifications and 13 different cancer types when Resnet50 is used as the feature extractor. Compared with the state-of-the-art methods focusing on 4–5 levels of magnification classification, the Hagnifinder achieves the best and most comparable performance in the BreakHis and Hagni40 datasets. ConclusionsThe experimental results suggest that Hagnifinder can be a valuable tool for predicting the associated magnification of any given histology image.

Models including preoperative plasma levels of angiogenic factors, leptin and IL-8 as potential biomarkers of endometrial cancer.

The diversity of endometrial cancer (EC) dictates the need for precise early diagnosis and pre-operative stratification to select treatment options appropriately. Non-invasive biomarkers invaluably assist clinicians in managing patients in daily clinical practice. Currently, there are no validated diagnostic or prognostic biomarkers for EC that could accurately predict the presence and extent of the disease. Our study analyzed 202 patients, of whom 91 were diagnosed with EC and 111 were control patients with the benign gynecological disease. Using Luminex xMAP™ multiplexing technology, we measured the pre-operative plasma concentrations of six previously selected angiogenic factors - leptin, IL-8, sTie-2, follistatin, neuropilin-1, and G-CSF. Besides basic statistical methods, we used a machine-learning algorithm to create a robust diagnostic model based on the plasma concentration of tested angiogenic factors. The plasma levels of leptin were significantly higher in EC patients than in control patients. Leptin was higher in type 1 EC patients versus control patients, and IL-8 was higher in type 2 EC versus control patients, particularly in poorly differentiated endometrioid EC grade 3. IL-8 plasma levels were significantly higher in EC patients with lymphovascular or myometrial invasion. Among univariate models, the model based on leptin reached the best results on both training and test datasets. A combination of age, IL-8, leptin and G-CSF was determined as the most important feature for the multivariate model, with ROC AUC 0.94 on training and 0.81 on the test dataset. The model utilizing a combination of all six AFs, BMI and age reached a ROC AUC of 0.89 on both the training and test dataset, strongly indicating the capability for predicting the risk of EC even on unseen data. According to our results, measuring plasma concentrations of angiogenic factors could, provided they are confirmed in a multicentre validation study, represent an important supplementary diagnostic tool for early detection and prognostic characterization of EC, which could guide the decision-making regarding the extent of treatment.

Identification and validation of a novel tumor driver gene signature for diagnosis and prognosis of head and neck squamous cell carcinoma.

Objective: Head and neck squamous cell carcinoma (HNSCC) is a common heterogeneous cancer with complex carcinogenic factors. However, the current TNM staging criteria to judge its severity to formulate treatment plans and evaluate the prognosis are particularly weak. Therefore, a robust diagnostic model capable of accurately diagnosing and predicting HNSCC should be established. Methods: Gene expression and clinical data were retrieved from The Cancer Genome Atlas and Gene Expression Omnibus databases. Key prognostic genes associated with HNSCC were screened with the weighted gene co-expression network analysis and least absolute shrinkage and selection operator (LASSO) Cox regression model analysis. We used the timeROC and survival R packages to conduct time-dependent receiver operating characteristic curve analyses and calculated the area under the curve at different time points of model prediction. Patients in the training and validation groups were divided into high- and low-risk subgroups, and Kaplan-Meier (K-M) survival curves were plotted for all subgroups. Subsequently, LASSO and support vector machine algorithms were used to screen genes to construct diagnostic model. Furthermore, we used the Wilcoxon signed-rank test to compare the half-maximal inhibitory concentrations of common chemotherapy drugs among patients in different risk groups. Finally, the expression levels of eight genes were measured using quantitative real-time polymerase chain reaction and immunohistochemistry. Results: Ten genes (SSB, PFKP, NAT10, PCDH9, SHANK2, PAX8, CELSR3, DCLRE1C, MAP2K7, and ODF4) with prognostic potential were identified, and a risk score was derived accordingly. Patients were divided into high- and low-risk groups based on the median risk score. The K-M survival curves confirmed that patients with high scores had significantly worse overall survival. Receiver operating characteristic curves proved that the prognostic signature had good sensitivity and specificity for predicting the prognosis of patients with HNSCC. Univariate and multivariate Cox regression analyses confirmed that the gene signature was an independent prognostic risk factor for HNSCC. Diagnostic model was built by identifying eight genes (SSB, PFKP, NAT10, PCDH9, CELSR3, DCLRE1C, MAP2K7, and ODF4). The high-risk group showed higher sensitivity to various common chemotherapeutic drugs. DCLRE1C expression was higher in normal tissues than in HNSCC tissues. Conclusion: Our study identified the important role of tumor-driver genes in HNSCC and their potential clinical diagnostic and prognostic values to facilitate individualized management of patients with HNSCC.

A machine learning model for predicting patients with major depressive disorder: A study based on transcriptomic data.

BackgroundIdentifying new biomarkers of major depressive disorder (MDD) would be of great significance for its early diagnosis and treatment. Herein, we constructed a diagnostic model of MDD using machine learning methods.MethodsThe GSE98793 and GSE19738 datasets were obtained from the Gene Expression Omnibus database, and the limma R package was used to analyze differentially expressed genes (DEGs) in MDD patients. Gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to identify potential molecular functions and pathways. A protein-protein interaction network (PPI) was constructed, and hub genes were predicted. Random forest (RF) and artificial neural network (ANN) machine-learning algorithms were used to select variables and construct a robust diagnostic model.ResultsA total of 721 DEGs were identified in peripheral blood samples of patients with MDD. GO and KEGG analyses revealed that the DEGs were mainly enriched in cytokines, defense responses to viruses, responses to biotic stimuli, immune effector processes, responses to external biotic stimuli, and immune systems. A PPI network was constructed, and CytoHubba plugins were used to screen hub genes. Furthermore, a robust diagnostic model was established using a RF and ANN algorithm with an area under the curve of 0.757 for the training model and 0.685 for the test cohort.ConclusionWe analyzed potential driver genes in patients with MDD and built a potential diagnostic model as an adjunct tool to assist psychiatrists in the clinical diagnosis and treatment of MDD.