Higher AUC Values Research Articles

Relationship of monocyte to high-density lipoprotein ratio (MHR) and other inflammatory biomarkers with sarcopenia: a population-based study

ObjectivesIn previous studies, several inflammatory biomarkers derived from complete blood cell counts (CBC), such as systemic immune inflammation index (SII), neutrophil-to-lymphocyte ratio (NLR), and non‑high‑density lipoprotein cholesterol to high‑density lipoprotein cholesterol ratio (NHHR) have been identified as predictors of sarcopenia. However, whether Monocyte to High-Density Lipoprotein Cholesterol Ratio (MHR) can predict the development of sarcopenia has not yet been established. The research first attempts to investigate the association between MHR and low muscle mass and to compare the predictive abilities of MHR, SII, NLR, and NHHR for low muscle mass risk.MethodsThe study comprised 10,321 participants aged 20 years and above from the United States. Multiple logistic regression was performed to explore the association between ln-transformed MHR, SII, NLR, NHHR and low muscle mass. Additionally, AUC values and ROC curves were used to assess the predictive effectiveness of ln MHR and other markers (ln SII, ln NLR, ln NHHR, ln MHR + ln SII, ln MHR + ln NHHR, and ln MHR + ln NLR). The bootstrap estimated 95% Cl was shown with the AUC.ResultsIn the fully adjusted model, ln SII, ln NLR, ln NHHR, ln MHR, ln MHR + ln SII, ln MHR + ln NHHR, and ln MHR + ln NLR were positively associated with low muscle mass (ln SII: OR = 1.59 [1.37–1.84]; ln NLR: OR = 1.35 [1.13–1.60]; ln NHHR: OR = 1.49[1.27–1.75]; ln MHR: OR = 1.98 [1.68–2.33]; ln MHR + ln SII: OR = 1.61 [1.46–1.79]; ln MHR + ln NHHR: OR = 1.42 [1.29–1.56]; ln MHR + ln NLR: OR = 1.58 [1.41–1.78]). Compared to the lowest quartile of ln MHR, higher quartiles were significantly associated with increased odds of low muscle mass (P for trend < 0.0001). In ROC analysis, ln MHR + ln SII had a higher AUC value than other indicators (AUC = 0.608).ConclusionLn-transformed MHR, SII, NLR, and NHHR were positively associated with low muscle mass. MHR outperforms SII, NLR, and NHHR in predicting sarcopenia.

Identifying corridors for Asiatic black bear (Ursus thibetanus) in a part of Eastern Himalayas, India

The Asiatic black bear (Ursus thibetanus), classified as a vulnerable species on the IUCN Red List, is an important mammal species found in the state of Sikkim, India. Studies carried out in Khangchendzonga National Park have documented the presence of these bears, highlighting their crucial conservation importance in the region. The population of Black bears are restricted to small habitat patches, which over the years have become fragmented by road networks and urban settlements. In such fragmented landscapes, connecting corridors play a crucial role in maintaining wildlife movement and genetic diversity. We assessed connectivity between eight protected areas in Sikkim using MaxENT and Circuitscape. 65 black bear presence locations (collected through Camera traps and sign surveys) and 24 environmental variables were used to model the corridors. Habitat suitability map was generated through MaxENT modelling approach. Our analysis suggests that there are multiple options to maintain connectivity for black bears in Sikkim. We mapped seven corridors and five pinch points (bottlenecks in connectivity), and calculated metrics to estimate their quality and importance. Our model output was supported by high AUC value (0.921) and field validation by questionnaire surveys and sign surveys to assess black bear presence and habitat use. Our results showed that 300 km² of the suitable regions are within the protected areas in Sikkim. The highest quality linkages as measured by the ratio of cost-weighted distance to Euclidean distance (CWD:EucD) and cost-weighted distance to least-cost path (CWD:LCP) were Khangchendzonga and Barsey, suggesting that these protected areas (National Parks and Wildlife Sanctuaries) and the developed corridors play important role in maintaining connectivity. We mapped pinch-points which are habitat where black bear movement is restricted due to unfavorable environments, linear infrastructures, built up/settlements or a combination of factors and our model predicted pinch points near few settlement areas; Mangan, Dikchu, Pangthang, Kabi, Yuksum and Lachen. Ground truthing confirmed that these areas also coincide with Black bear conflict zones in Sikkim.

Development a glycosylated extracellular vesicle-derived miRNA Signature for early detection of esophageal squamous cell carcinoma

BackgroundEsophageal squamous cell carcinoma (ESCC) is often diagnosed at an advanced stage due to the lack of non-invasive early detection tools, which significantly impacts patient prognosis. Given that glycosylation alterations especially high sialylation and fucosylation, frequently occur during cellular malignant transformation, but their roles are not elucidated. We examined alterations in disease-specific glycosylated extracellular vesicles (EVs)-derived miRNAs in the serum of ESCC patients, evaluating their utility as diagnostic biomarkers.MethodsA total of 371 ESCC and 303 healthy controls (HCs) were recruited in this multi-stage, multicentre case-control study. Fucosylated (Fuc-) and sialylated (Sia-) EVs were isolated utilizing Lentil lectin (LCA) and wheat germ lectin (WGA)-coated magnetic beads, respectively. The glycosylated EVs-derived miRNAs-based signature (RiskscoreFuc−&Sia−) was established through logistic regression in a training cohort and subsequently validated in an internal and an external multicentre cohort.ResultsThe RiskscoreFuc−&Sia− effectively identified ESCC across all stages, demonstrating high AUC values in training (0.980), internal validation (0.957), and external multicentre validation (0.973) cohorts, markedly higher than carcinoembryonic antigen (CEA) (AUC = 0.769, training cohort; AUC = 0.749, internal validation cohort; AUC = 0.765, external validation cohort). Notably, this score exhibited robust accuracy in detecting CEA (-) ESCC cases (CEA < 5 ng/ml) (AUC = 0.974, training & internal cohort; AUC = 0.973, external multicentre validation cohort). Additionally, it displayed strong efficacy in differentiating early-stage ESCC patients (AUC = 0.982, training cohort; AUC = 0.977, external multicentre validation cohort).ConclusionsOur study illustrates the effectiveness of glycosylated EVs capture strategy for isolating tumour-specific EVs. The unique glycosylated EVs-derived miRNAs-based signature shows the optimal potential as a biomarker for early detection of ESCC.Graphical

Integrating a multi-dimensional deep convolutional neural network with optimized sample selection for landslide susceptibility assessment

ABSTRACT To address the errors of negative samples in landslide susceptibility modeling and traditional methods in exploring the regularities hidden in the evaluation factors, this paper proposes a stacking one- and three-dimensional Convolutional Neural Network (Stacking-1D-3D-CNN) landslide susceptibility assessment method considering sample optimization selection. First, in order to select negative samples rationally, this paper adopts the Relative Frequency Ratio combined with Certainty Factor Method (RFR-CFM) to determine the negative samples; secondly, the Stacking-1D-3D-CNN proposed is combined with RFR-CFM for the first time for landslide susceptibility assessment. In this work, the negative samples determined by RFR-CFM and the Information Quality Model (IQM) were combined with historical disaster points to form a total modeling sample, and modeled at different ratios. Finally, it is compared with several other models in terms of the landslide hazard susceptibility zoning results, prone zone statistics, and model performance. The findings show that the degree of spatial aggregation of training samples and testing samples has a much greater impact on the accuracy of landslide susceptibility modeling than the impact of their proportions. Furthermore, compared with other models, RFR-CFM-Stacking-1D-3D-CNN has the highest AUC value, precision, recall, F-score, and accuracy, which are 0.95, 0.83, 0.89, 0.85, and 84.76%, respectively, and the lowest RMSE and MAE, 0.39 and 0.15, respectively. This proves the RFR-CFM sample selection method’s rationality and the Stacking-1D-3D-CNN model’s effectiveness.

Predicting habitat refugia of the medicinal plant Dolomiaea costus under two climate change scenarios

Assessing the impacts of climate change on the distribution of medicinally important and endangered species is critical for conservation planning. This study focuses on Dolomiaea costus (Falc.) Kasana & A.K. Pandey, a critically endangered and highly valuable medicinal plant found in the fragile climate of Pakistan. Using species distribution modelling (MaxEnt), we evaluated the current and future habitat suitability of D. costus under two climate change scenarios (SSP370 and SSP585) for the periods 2041–2070 and 2071–2100. Occurrence data were collected from field surveys and the Global Biodiversity Information Facility (GBIF), while bioclimatic data were sourced from Climatologies at High Resolution for the Earth’s Land Surface Areas (CHELSA). The MaxEnt model achieved a high mean AUC value of 0.967, indicating strong predictive accuracy. Suitable habitats for D. costus span approximately 99,871 Km2 across the western Himalayas, Karakoram, and Hindukush regions. Climate projections revealed that under SSP370, suitable areas decrease from 99,260 Km2 in 2041–2070 to 90,962.5 Km2 in 2071–2100, with high-suitability zones reducing from 13,024.5 Km2 to 11,490.5 Km2. In contrast, the SSP585 scenario showed a modest expansion of suitable habitat from 107,104 Km2 to 111,217.5 Km2, though high-suitability areas declined from 18,208 Km2 to 16,811 Km2. Precipitation-related variables, especially precipitation seasonality, were the most influential drivers of habitat suitability, contributing 64.3% to the model, followed by mean monthly precipitation of the driest quarter (18.8%) and precipitation of the wettest and driest months. Jackknife tests also confirmed their critical importance. These findings underscored the sensitivity of D. costus habitat to the intensity of climate change, particularly under severe warming conditions and highlighted the urgent need for a comprehensive conservation plan. A combination of strategies, including climate-smart conservation, habitat protection, species reintroduction, and cultivation in moderately to highly suitable habitats, is crucial to ensure the long-term survival of D. costus in Pakistan amidst the changing climate.

Digital Profiling of Tumor Extracellular Vesicle-Associated RNAs Directly from Unprocessed Blood Plasma.

Tumor-derived extracellular vesicle (tEV)-associated RNAs hold promise as diagnostic biomarkers, but their clinical use is hindered by the rarity of tEVs among nontumor EVs. Here, we present EV-CLIP, a highly sensitive droplet-based digital method for profiling EV RNA. EV-CLIP utilizes the fusion of EVs with charged liposomes (CLIPs) in a microfluidic chip. Optimized CLIP surface charge enables exceptional sensitivity and selectivity for EV-derived miRNAs and mRNAs. This approach streamlines detection with minimal plasma volume (20 μL) and eliminates the need for prior EV isolation or RNA preparation, preventing loss of EVs or RNA. In testing with 83 patient samples, EV-CLIP detected EGFR L858R and T790M mutations with high AUC values of 1.0000 and 0.9784, respectively. Its success in serial monitoring during chemotherapy highlights its potential for precise quantification of rare EV subpopulations, facilitating the exploration of single EV RNA content and enhancing understanding of diverse EV populations in various disease states.

Identification and experimental validation of diagnostic and prognostic genes CX3CR1, PID1 and PTGDS in sepsis and ARDS using bulk and single-cell transcriptomic analysis and machine learning

BackgroundSepsis is an uncontrolled reaction to infection that causes severe organ dysfunction and is a primary cause of ARDS. Patients suffering both sepsis and ARDS have a poor prognosis and high mortality. However, the mechanisms behind their simultaneous occurrence are unclear.MethodsWe acquired sepsis and ARDS datasets from GEO and Arrayexpress databases and screened hub genes by WGCNA and machine learning algorithm. For diagnosis and prognosis, ROC curve and survival analysis were used. We performed GO, KEGG, GSEA, immune cell infiltration, drug prediction, molecular docking, transcription factor prediction, and constructed PPI and ceRNA networks to explore these genes and the common mechanisms of sepsis and ARDS. Single-cell data analysis compared immune cell profiles and hub gene localization. Finally, RT-qPCR and H&amp;amp;E staining confirmed the reliability of hub genes using PBMCs samples and mouse models.ResultsWe identified 242 common differentially expressed genes in sepsis and ARDS. WGCNA analysis showed that the turquoise module in GSE95233 is strongly linked to sepsis occurrence and poor prognosis, while the black module in GSE10474 is associated with ARDS. Using WGCNA and three machine learning methods (LASSO, random forest and Boruta), we identified three key genes CX3CR1, PID1 and PTGDS. Models built with them showed high AUC values in ROC curve evaluations and were validated by external datasets, accurately predicting the occurrence and mortality. We further explored the immunological landscape of these genes using immune infiltration and single-cell analysis. Then, the ceRNA, predicted drugs and molecular docking were analyzed. Ultimately, we demonstrated that these genes are expressed differently in human and mouse samples with sepsis and ARDS.ConclusionThis study identified three molecular signatures (CX3CR1, PID1 and PTGDS) linked to the diagnosis and poor prognosis of sepsis and ARDS, validated by RT-qPCR and H&amp;amp;E staining in both patient and mouse samples. This research may be valuable for identifying shared biological mechanisms and potential treatment targets for both diseases.

Does amide proton transfer-weighted MRI have diagnostic and differential value in ovarian cystic and predominantly cystic lesion?

This study aims to evaluate the diagnostic value of amide proton transfer-weighted (APTw) imaging in distinguishing cystic or predominantly cystic ovarian lesions. 49 patients underwent APTw imaging at 3T-MR before surgery, with 20 volunteers serving as the control group. Participants were divided into the following groups: solid components of normal ovaries (Group A, n = 29), solid components of malignant lesions (Group B, n = 7), cystic fluid of follicles (Group C, n = 31), cystic fluid of benign lesions (Group D, n = 46), functional cysts (Group d1, n = 8), endometriomas (Group d2, n = 28), cystadenomas (Group d3, n = 10), and cystic fluid of malignant lesions (Group E, n = 12). Independent t-tests or Mann-Whitney U tests and one-way ANOVA were used to compare group differences. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic efficacy in distinguishing between different lesions. For solid components, significant differences in MTRasym values were observed between Groups A and B (P < 0.001). For cystic components, significant differences were found between Groups C and D, C and E, d1 and d2, d2 and d3, d1 and d3, C and d2, C and d3, E and d1, and E and d2 (all P < 0.01). ROC analysis of these results showed high AUC values (ranging from 0.813 to 1.0), all P < 0.05. APTw can reveal differences in MTRasym values between normal and diseased ovarian tissues, demonstrating high clinical value in differentiating functional cysts, endometriomas, and cystadenomas, as well as distinguishing benign lesions (functional cysts or endometriomas) from malignant tumors.

Clinical value of contrast-enhanced ultrasound combined with quantitative analysis in Bosniak ≥ II cystic renal masses.

The 2019 Bosniak classification has improved the precise and detailed quantitative evaluation criteria, making the diagnosis of cystic renal masses (CRMs) more accurate and objective. This study addresses the clinical value of quantitative analysis and aims to investigate the feasibility of combining contrast-enhanced ultrasound (CEUS) with quantitative analysis for diagnosing Bosniak ≥ II CRMs. We retrospectively obtained 58 CRMs with confirmed pathology, which underwent CEUS and Contrast-enhanced computer tomography (CECT) evaluations according to Bosniak classification between January 2013 and August 2024. These lesions were divided into benign and malignant groups, followed by a quantitative analysis of the morphological details detected by CEUS. All morphological parameters were compared, and the diagnostic efficiencies were evaluated using receiver operating characteristics (ROC) curves, logistic regression (LR) analysis, and diagnostic curve analysis (DCA). Additionally, a cohort of 72 lesions was monitored for a period of ≥ 3years, and changes in Bosniak classification were analyzed by categorizing them into stable, upgraded, and downgraded categories. The analysis revealed no statistically significant difference between CEUS and CECT in our cohort's malignancy predictive rates across different Bosniak grades (p = 0.640). All morphological quantitative parameters showed statistically significant differences between the two groups (p < 0.001). ROC curve analysis revealed that the sum of enhanced wall thickness and enhanced septum thickness among quantitative parameters had the highest AUC value (AUC: 0.9226). Both LR models demonstrated superior clinical diagnostic performance with similar level of accuracy between qualitative and quantitative analysis, as evidenced by ROC (AUC: 0.9470, 0.9619, respectively) and DCA analyses. None of the lesions in the follow-up cohort were upgraded, suggesting that CRMs are relatively stable tumors with a low malignant potential. This retrospective study demonstrated that CEUS combined with Bosniak classification and quantitative analysis could enhance diagnostic confidence in differentiating Bosniak ≥ II CRMs and could serve as a viable alternative to CECT in specific cases.

BIRC5 as a prognostic and diagnostic biomarker in pan-cancer: an integrated analysis of expression, immune subtypes, and functional networks.

BIRC5 (Survivin) is a crucial anti-apoptotic protein overexpressed in various cancers, promoting tumor growth and treatment resistance. This study investigates its expression across 33 cancer types and explores its diagnostic, prognostic, and immune-related significance. We analyzed RNA-seq data from TCGA and protein expression data from the Human Protein Atlas. Expression levels were compared between tumor and normal tissues. Correlations with molecular and immune subtypes were explored using TISIDB. Prognostic significance was evaluated through survival analysis, Cox regression, and ROC curve analysis. The PPI network was constructed using STRING. BIRC5 was significantly overexpressed in tumor tissues across 33 cancer types, with higher expression levels observed in tumors compared to normal tissues. The protein expression analysis revealed a similar trend. BIRC5 expression was significantly correlated with various molecular and immune subtypes in multiple cancer types. Survival analysis indicated that high BIRC5 expression was associated with poor prognosis across multiple cancers, including lung adenocarcinoma (LUAD) and kidney renal clear cell carcinoma (KIRC). ROC analysis showed that BIRC5 exhibited strong diagnostic potential, with high AUC values (>0.9) in several cancers. The PPI network analysis identified key interacting proteins involved in the cell cycle and tumor progression, further supporting BIRC5's role in cancer biology. Functional experiments in lung adenocarcinoma (LUAD) revealed that BIRC5 upregulation enhances cell proliferation, migration, and invasion, while its knockdown suppresses these activities. BIRC5 is a promising diagnostic and prognostic biomarker in multiple cancers. Its association with immune subtypes suggests a potential role in the tumor immune microenvironment. These findings support BIRC5 as a therapeutic target for cancer treatment.

Exploring the impact of deubiquitination on melanoma prognosis through single-cell RNA sequencing.

Cutaneous melanoma, characterized by the malignant proliferation of melanocytes, exhibits high invasiveness and metastatic potential. Thus, identifying novel prognostic biomarkers and therapeutic targets is essential. We utilized single-cell RNA sequencing data (GSE215120) from the Gene Expression Omnibus (GEO) database, preprocessing it with the Seurat package. Dimensionality reduction and clustering were executed through Principal Component Analysis (PCA) and Uniform Manifold Approximation and Projection (UMAP). Cell types were annotated based on known marker genes, and the AUCell algorithm assessed the enrichment of deubiquitination-related genes. Cells were categorized into DUB_high and DUB_low groups based on AUCell scores, followed by differential expression analysis. Importantly, we constructed a robust prognostic model utilizing various genes, which was evaluated in the TCGA cohort and an external validation cohort. Our prognostic model, developed using Random Survival Forest (RSF) and Ridge Regression methods, demonstrated excellent predictive performance, evidenced by high C-index and AUC values across multiple cohorts. Furthermore, analyses of immune cell infiltration and tumor microenvironment scores revealed significant differences in immune cell distribution and microenvironment characteristics between high-risk and low-risk groups. Functional experiments indicated that TBC1D16 significantly impacts the migration and proliferation of melanoma cells. This study highlights the critical role of deubiquitination in melanoma and presents a novel prognostic model that effectively stratifies patient risk. The model's strong predictive ability enhances clinical decision-making and provides a framework for future studies on the therapeutic potential of deubiquitination mechanisms in melanoma progression. Further validation and exploration of this model's applicability in clinical settings are warranted.

Aspartate aminotransferase-to-platelet ratio index outperforms Fibrosis-4 in 2843 Korean patients with metabolic dysfunction-associated steatotic liver disease.

The definition of metabolic dysfunction-associated steatotic liver disease (MASLD) has recently been proposed. We aim to investigate the diagnostic efficacy of noninvasive fibrosis markers in predicting liver fibrosis in patients with nonalcoholic fatty liver disease (NAFLD), metabolic dysfunction-associated fatty liver disease (MAFLD), and MASLD. This retrospective study involved 2843 patients diagnosed with steatotic liver disease at six tertiary hospitals in South Korea. Liver fibrosis was assessed using vibration-controlled transient elastography, and various noninvasive markers, including the aspartate aminotransferase-to-platelet ratio index (APRI), Fibrosis-4 index (FIB-4), NAFLD fibrosis score (NFS), and serum Mac-2-binding protein glycosylation isomer were analyzed. Among 1106 patients, 79.9% met criteria for NAFLD, MAFLD, and MASLD. The APRI had area under the receiver operating characteristic curve (AUC) values of 0.819, 0.821, and 0.818 for liver fibrosis≥F2, and 0.819, 0.824, and 0.884 for liver fibrosis≥F3, and 0.890, 0.884, and 0.889 for fibrosis≥F4 in NAFLD, MAFLD, and MASLD, respectively. The FIB-4 index showed AUC values of 0.776, 0.793, and 0.778 for fibrosis≥F2, 0.788, 0.814, and 0.79 for fibrosis≥F3, and 0.846, 0.859, and 0.856 for fibrosis≥F4. The APRI consistently had the highest AUC values, except in individuals older than 64years for fibrosis≥F4. The APRI was the most effective noninvasive fibrosis marker across NAFLD, MAFLD, and MASLD, particularly in age-stratified analyses. Further research is needed to establish standardized cut-off values and enhance the clinical utility of these markers in managing liver fibrosis.

Predicting Habitat Suitability for Tarantula in Peninsular Malaysia by Using Species Distribution Modelling (SDM)

Theraphosidae, usually referred to as "Tarantula," is an infraorder of the Mygalomorh arachnid family of spiders. It is a large spider with downward-pointing fangs and a hairy body. As of now, few studies were done on Theraphosidaes in Peninsular Malaysia, but those were mainly concentrated on physical taxonomy with brief explanations of the spiders’ natural history. This study revealed the prediction of Theraphosidae habitat suitability in Peninsular Malaysia. Species Distribution Modelling (SDM) was used based on Theraphosidae points of occurrence in Peninsular Malaysia. Prediction for habitat suitability was determined from 274 coordinates of occurrence. In the study, a high AUC value of 0.857 indicated a high discrimination power of the predictive model. Tree canopy cover, elevation and forested area were the top three significant factors that could predict the distribution of Theraphosidae species, according to the SDM modelling by using MaxENT. Based on this species distribution modelling study, it was concluded that protecting the hilly forested area in Malaysia was crucial for the protection of Theraphosidae spiders. The study highlighted the influence of environmental factors on Theraphosidae distributions in Peninsular Malaysia and provided insights for future research and conservation efforts.

Machine learning for epithelial ovarian cancer platinum resistance recurrence identification using routine clinical data.

Most epithelial ovarian cancer (EOC) eventually develops recurrence. Identification of high-risk patients can prompt earlier intervention and improve long-term outcomes. We used laboratory and clinical data to create models based on machine learning for EOC platinum resistance recurrence identification. This study was designed as a retrospective cohort analysis. Initially, we identified 1,392 patients diagnosed with epithelial ovarian cancer who underwent platinum-based chemotherapy at Yunnan Cancer Hospital between January 1, 2012, and June 30, 2022. We collected data on the patients' clinicopathologic characteristics, routine laboratory results, surgical information, details of chemotherapy regimens, and survival outcomes. Subsequently, to identify relevant variables influencing the recurrence of platinum resistance, we screened thirty potential factors using two distinct variable selection methods: Lasso regression and multiple logistic regression analysis. Following this screening process, five machine learning algorithms were employed to develop predictive models based on the selected variables. These included decision tree analysis (DTA), K-Nearest neighbor (KNN), support vector machine (SVM), random forest (RF), and eXtreme gradient boosting (XGBoost). The performance of these models was compared against that of traditional logistic regression. To ensure robust internal validation and facilitate comparison among model performance metrics, a five-fold cross-validation method was implemented. Key performance indicators for the models included the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and average accuracy. Finally, we will visualize these models through nomograms, decision tree diagrams, variable importance plots, etc., to assist clinicians in their practice. Multiple logistic regression analysis identified eight variables associated with platinum resistance recurrence. In the lasso regression, seven variables were selected. Based on the findings from both Lasso regression and multiple logistic regression analysis, models were developed using these 7 and 8 factors. Among these, the XGBoost model derived from multiple logistic regression exhibited superior performance and demonstrated good discrimination during internal validation, achieving an AUC of 0.784, a sensitivity of 0.735, a specificity of 0.713, an average accuracy of 80.4%, with a cut-off value set at 0.240. Conversely, the LR model based on lasso regression yielded commendable results as well; it achieved an AUC of 0.738, a sensitivity of 0.541, a specificity of 0.836, with a cut-off value established at 0.154 and an accuracy rate of 79.6%. Finally, we visualized both models through nomograms to illustrate the significance of each variable involved in their development. We have successfully developed predictive models for platinum-resistant recurrence of epithelial ovarian cancer, utilizing routine clinical and laboratory data. Among these models, the XGBoost model-derived from variables selected through multiple logistic regression-demonstrated the best performance. It exhibited high AUC values and average accuracy during internal validation, making it a recommended tool for clinical use. However, due to variations in time and context, influencing factors may change over time; thus, continuous evolution of the model is necessary. We propose a framework for this ongoing model adaptation.

Implementation of PPCA Imputation, SMOTE-N Class Balancing in Hepatitis Classification Using Naïve Bayes

The availability of complete data in research is crucial, especially in the initial stages. The Hepatitis data used in this study encountered issues such as missing data and class imbalance, which hindered its optimal utilization. The method employed to address missing data was the PPCA imputation method. After filling in the missing data, the data was balanced using the SMOTE-N class balancing method and classified using Gaussian Naïve Bayes. The aim of this research was to compare the classification evaluation of hepatitis disease using Naive Bayes with the PPCA imputation approach and SMOTE-N class balancing. The best results from each scenario yielded an AUC value of 0.833 in the first scenario with an 80:20 data split for training and testing, and 0.875 in the second scenario with a 90:10 data split. The highest AUC value was obtained in the application of PPCA imputation with SMOTE-N class balancing using Naive Bayes classification. This demonstrates that the implementation of PPCA imputation with SMOTE-N class balancing has a better impact on the performance of Naïve Bayes classification.

Fabrication and Optimization of a Silodosin In Situ-Forming PLGA Implants for the Treatment of Benign Prostatic Hyperplasia: In Vitro and In Vivo Study.

Objectives: Lower urinary tract symptoms (LUTSs) related to benign prostatic hyperplasia (BPH) are common in older men, and alpha-adrenoceptor blockers continue to be a key part of managing these symptoms. This study aimed to formulate injectable poly (lactic-co-glycolic acid) (PLGA) in situ-forming implants (ISFIs) loaded with silodosin (SLD) to address symptoms associated with BPH. This method, which ensures prolonged therapeutic effects of SLD, is intended to decrease dosing frequency and improve treatment outcomes, leading to better patient adherence. Methods: An appropriate solvent with favorable PLGA solubility, viscosity, and in vitro release profile was selected. Additionally, an I-optimal design was employed as an optimization technique. An in vivo study in albino male rats was conducted to investigate prostate-specific antigens (PSAs), prostate weight and prostatic index, histopathology, and SLD pharmacokinetics. Results: The optimized formulation showed experimental values of 29.25% for the initial burst after 2 h and 58.23% for the cumulative release of SLD after 10 days. Pharmacokinetic data revealed that the SLD-ISFI formulation had lower Cmax and higher AUC values than subcutaneous (SC) pure SLD and oral commercial SLD capsule, indicating the controlled-release impact and improved bioavailability of the ISFI systems. SLD-ISFI produced a marked drop in the prostatic index by 2.09-fold compared to the positive control. Serum PSA level decreased significantly from 0.345 ± 0.007 to 0.145 ± 0.015 ng/mL after SLD-ISFI injection compared to the positive control. Conclusions: This study indicated that the optimized SLD-ISFI formulation proved its efficacy in managing BPH.

An optimized method for dose-effect prediction of traditional Chinese medicine based on 1D-ResCNN-PLS.

We introduce a one-dimensional (1D) residual convolutional neural network with Partial Least Squares (1D-ResCNN-PLS) to solve the covariance and nonlinearity problems in traditional Chinese medicine dose-effect relationship data. The model combines a 1D convolutional layer with a residual block to extract nonlinear features and employs PLS for prediction. Tested on the Ma Xing Shi Gan Decoction datasets, the model significantly outperformed conventional models, achieving high accuracies, sensitivities, specificities, and AUC values, with considerable reductions in mean square error. Our results confirm its effectiveness in nonlinear data processing and demonstrate potential for broader application across public datasets.

Study on the feasibility of distinguishing ventricular and pre-excited arrhythmia rhythms by a new algorithm

BackgroundThe differentiation and diagnosis of ventricular tachycardia (VT) and pre-excited tachycardia (PXT) remains a challenging task, especially when typical AV dissociation is not present. The purpose of this article is to study the feasibility of a new theoretical algorithm for identifying ventricular arrhythmias (VA) and pre-excited arrhythmias (PA) rhythms (which can be used to distinguish VT from PXT, etc.). MethodThis study involved the deduction of a new algorithm by combining knowledge of cardiac anatomy, vectorcardiography, and cardiac electrophysiology. The new algorithm evaluated the diagnostic value through intracardiac electrophysiology in 205 cases of VA and PA. The new algorithm diagnoses VA based on the following 4-step process:1.The QRS complex in leads II, III, and aVF shows a unidirectional R wave, and lead aVR shows a QS pattern.2.S waves are predominant in two or more of leads I, aVF, and V6.3.Lead V2 shows ≥3 phase waves or returning branch notching (note: returning branch refers to the band of QRS complexes returning to the baseline).4.Lead V5 shows a negative wave in the initial portion or returning branch notching.If none of these criteria are met, the diagnosis is PA. The diagnostic value of the new algorithm is compared with the Steurer algorithm and the Vereckei algorithm (diagnosed based on the QRS waveform characteristics of the two algorithms during electrophysiological verification, excluding the diagnosis of atrioventricular dissociation). ResultsThe new algorithm showed significant advantages in terms of AUC value (0.83 vs. 0.61 vs. 0.57), sensitivity (83.6 % vs. 23.3 % vs. 24.8 %), and accuracy (82.9 % vs. 48.3 % vs. 46.3 %) compared to the Steurer algorithm and Vereckei algorithm based on QRS waveform characteristics for diagnosing VA (137 cases) and PA (68 cases). This indicates that the new algorithm is more accurate in identifying idiopathic VA. While there was a significant difference in specificity between the New algorithm and Steurer algorithm (82.3 % vs. 98.5 %, p < 0.05), the difference with Vereckei algorithm (82.3 % vs. 89.7 %) was not significant.In the New algorithm, the sensitivity and specificity for each step are as follows:-Step 1: Sensitivity 34.3 %, Specificity 94.1 %.-Step 2: Sensitivity 24.1 %, Specificity 98.5 %.-Step 3: Sensitivity 18.3 %, Specificity 100 %.-Step 4: Sensitivity 6.6 %, Specificity 89.7 %.Step 1 had the highest AUC value, indicating the best overall diagnostic performance among all steps. Step 2 and Step 3 also performed well, while Step 4 had relatively poorer diagnostic performance. ConclusionThe new algorithm is suitable for identifying the origin of VA and PA rhythms.

The incremental value of interatrial septum motion in predicting thrombus or spontaneous echo contrast in patients with non-valvular atrial fibrillation: an observational study on transesophageal echocardiography.

The thickness and motion of the Interatrial Septum (IAS) possibly serves as indicators of both structural and functional remodeling of left atrium. This study aims to use transesophageal echocardiography (TEE) to assess IAS motion in non-valvular atrial fibrillation (NVAF) and investigate its correlation with the risk of spontaneous echo contrast (SEC) and thrombus (TH). We conducted a cross-sectional study on 318 patients with NVAF who underwent transthoracic echocardiography and TEE. IAS motion was defined as the maximum displacement of IAS observed throughout the cardiac cycles using M-mode TEE. The prevalence of SEC/TH was 39.9% (127/318) in the overall group. In paroxysmal atrial fibrillation (PAF), the prevalence rate of SEC/TH was 25.3% (47/186), compared to 60.6% (80/132) in persistent atrial fibrillation (PeAF). Multivariable analyses showed that LA volume index (LAVI), mean E/e', S/D ratio, IAS motion and CHA2DS2-VASc scores were significantly associated with SEC/TH. Patients with lower IAS motion showed a higher prevalence of SEC/TH compared to those with higher IAS motion (71.3% vs. 11.6%; P < 0.001). The IAS motion under sinus rhythm could better predict risk of SEC/TH, while the predictive efficacy under AF rhythm was slightly attenuated but still had a high AUC value (0.779). A significant positive correlation was observed between the IAS motion and the LAA filling velocity (PAF r = 0.47; P<0.001 and PeAF r = 0.38; P < 0.001, respectively), LAA emptying velocity (PAF r = 0.55; P < 0.001 and PeAF r = 0.47; P < 0.001, respectively) and LAVI (PAF r = 0.59; P < 0.001 and PeAF r = 0.44; P < 0.001, respectively). The integration of the IAS motion to the CHA2DS2-VASc, LAVI and mean E/e' provided important incremental predictive value of SEC/TH (AUC = 0.859 vs. 0.826, P = 0.02). IAS motion measured by TEE correlates well with LAA flow velocity and LA size and is independently associated with SEC/TH in NVAF. Additionally, lower IAS motion is associated with a higher prevalence of SEC/TH. Furthermore, the integration of IAS motion to CHA2DS2-VASc, LAVI and mean E/e' can provide additional value for the prediction of SEC/TH.

Predicting the Risk of Loneliness in Children and Adolescents: A Machine Learning Study.

Loneliness is increasingly emerging as a significant public health problem in children and adolescents. Predicting loneliness and finding its risk factors in children and adolescents is lacking and necessary, and would greatly help determine intervention actions. This study aimed to find appropriate machine learning techniques to predict loneliness and its associated risk factors among schoolchildren. The data were collected from an ongoing prospective puberty cohort that was established in Chongqing, Southwest China. This study used 822 subjects (46.84% boys, age range: 11-16) followed in 2019. Five models, (a) random forest, (b) extreme gradient boosting (XGBoost), (c) logistic regression, (d) neural network, and (e) support vector machine were applied to predict loneliness. A total of 39 indicators were collected and 28 predictors were finally included for prediction after data pre-processing, including demographic, parental relationship, mental health, pubertal development, behaviors, and environmental factors. Model performance was determined by accuracy and AUC. Additionally, random forest and XGBoost were applied to identify the important factors. The XGBoost algorithm with SHAP was also used to interpret the results of our ML model. All machine learning performed with favorable accuracy. Compared to random forest (AUC: 0.87 (95%CI: 0.80, 0.93)), logistic regression (AUC: 0.80 (95%CI: 0.70, 0.89)), neural network (AUC: 0.80 (95%CI: 0.71, 0.89)), and support vector machine (AUC: 0.79 (95%CI: 0.79, 0.89)), XGBoost algorithm had the highest AUC values 0.87 (95%CI: 0.80, 0.93) in the test set, although the difference was not significant between models. Peer communication, index of general affect, peer alienation, and internet addiction were the top four significant factors of loneliness in children and adolescents. The results of this study suggest that machine learning has considerable potential to predict loneliness in children. This may be valuable for the early identification and intervention of loneliness.