Area Under Curve Values Research Articles

DoFA: Adversarial examples detection for SAR images by dual-objective feature attribution

Synthetic aperture radar (SAR) classification models based on convolutional neural networks have high accuracy, but the models’ security is still threatened by adversarial examples. The high threat of adversarial examples derives from the invisible noise that can cause feature changes within the model. Among many adversarial examples detection methods, feature attribution that is sensitive to feature changes performs well in feature analysis. Unfortunately, the existing feature attribution-based detection methods cannot balance the computational efficiency and detection performance well due to the size and speckle noise of the images in SAR adversarial examples detection. In this work, we propose the Dual-objective Feature Attribution (DoFA) method by using the feature attribution scan block to find the suitable scan granularity. The DoFA method formulates the SAR adversarial examples detection issue as a dual-objective optimization problem and takes the number of subsamples generated by feature analysis and the area under curve (AUC) value of the logistic regression model as the objective functions while the feature scan block’s size, stride, padding, and the number of selected model layers are the decision variables. Non-dominated Sorting Genetic Algorithm-II (NSGA-II) is adopted to search for the scan block parameters with Pareto optimality so that the DoFA method can automatically obtain the best feature analysis granularity in different scenarios. The experimental results on the FUSAR-Ship dataset have shown that the proposed DoFA method has a higher AUC value under five adversarial attacks and a smaller number of subsamples than the existing adversarial examples detection method.

Spatial Mapping for Multi-Hazard Land Management in Sparsely Vegetated Watersheds Using Machine Learning Algorithms

This study breaks new ground by developing a multi-hazard vulnerability map for the Tensift watershed and the Haouz plain in the Moroccan High Atlas area. The unique juxtaposition of flat and mountainous terrain in this area increases sensitivity to natural hazards, making it an ideal location for this research. Previous extreme events in this region have underscored the urgent need for proactive mitigation strategies, especially as these hazards increasingly intersect with human activities, including agriculture and infrastructure development. In this study six advanced machine learning (ML) models were used to comprehensively assess the combined probability of three significant natural hazards: flooding, gully erosion, and landslides. These models rely on causal factors derived from reputable sources, including geology, topography, meteorology, human activities, and hydrology. The research's rigorous validation process, which includes metrics such as specificity, precision, sensitivity, and accuracy, underlines the robust performance of all six models. The validation process involved comparing the model's predictions with actual hazard occurrences over a specific period. According to the outcomes in terms of the area under curve (AUC), the XGBoost model emerged as the most predictive, with remarkable AUC values of 93.41% for landslides, 91.07% for gully erosion and 93.78% for flooding. Based on the overall findings of this study, a multi-hazard risk map was created using the relationship between flood risk, gully erosion, and landslides in a geographic information system (GIS) architecture. The innovative approach presented in this work, which combined ML algorithms with geographical data, demonstrates the power of these tools in sustainable land management and the protection of communities and their assets in the Moroccan High Atlas and regions with similar topographical, geological, and meteorological conditions that are vulnerable to the aforementioned risks.

The Factors for the Occurrence of Pulmonary Infection after Gastrointestinal Surgery and the Construction of a Predictive Model Using sTREM-1 and TIM-4: A Retrospective Study.

Identifying and intervening with high-risk postoperative pulmonary infections patients pose challenges in clinical practice. This study aims to conduct a comprehensively analysis of the risk factors and predictive factors associated with post-gastrointestinal surgery pulmonary infections and to develop a predictive model that can predict occurrence of pulmonary infection. A retrospective analysis was conducted on 96 patients who underwent gastrointestinal surgery at our hospital from May 2021 to October 2023. The occurrence rate of postoperative pulmonary infections was calculated, and patients were categorized into two groups: those with pulmonary infections (the occurrence group) and those without pulmonary infections (the non-occurrence group). Logistic regression analysis was utilized to identify the risk factors for post-gastrointestinal surgery pulmonary infections and to evaluate the predictive value of soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) and T cell immunoglobulin and mucin domain-4 (TIM-4) using nomograms, calibration curves, and Receiver Operating Characteristic (ROC) curves. Out of 96 patients, 20 (20.83%) developed postoperative pulmonary infections. Significant differences were noted between occurrence and non-occurrence groups in terms of smoking (65.00% vs. 34.21%, p = 0.013), surgical duration (70.00% vs. 31.58%, p = 0.002), Preoperative hemoglobin level (35.00% vs. 65.79%, p = 0.013), sTREM-1 levels (23.57 ± 3.16 pg/mL vs. 15.62 ± 2.48 pg/mL, p < 0.001), and TIM-4 levels (61.48 ± 6.35 pg/mL vs. 44.73 ± 5.22 pg/mL, p < 0.001). Logistic regression analysis leads to the development of a risk prediction model for post-gastrointestinal surgery pulmonary infections. The high predictive values of sTREM-1 (Area Under Curve (AUC) = 0.962, 95% confidence interval (CI) 0.917~0.999) and TIM-4 (AUC = 0.970, 95% CI 0.925~1.000) were highlighted by the AUC values, underscoring their clinical importance. A predictive model utilizing sTREM-1 and TIM-4 for pulmonary infection following gastrointestinal surgery was developed. Additionally, other risk factors such as smoking, surgical duration, and preoperative hemoglobin level were evaluated. This finding can be applied in clinical practice to identify potentially susceptible patients and facilitate early intervention measures.

Divorce Factor Classification Uses The C4.5 Algorithm Based On Particle Swarm Optimization

Cases of household divorce increased in the West Java area during the Covid-19 pandemic. The pandemic has increased personal relationships and interactions between family members, and some families are using this opportunity to strengthen their relationships. However, increased family interaction can also result in increased conflict, leading to divorce. The author classifies divorce factors that have increased during the pandemic using the C4.5 Algorithm based on Particle Swarm Optimization (PSO). The main factors for divorce are economic factors that have hit during the pandemic coupled with unstable mental conditions resulting in poor communication and continuous fighting. So that the husband/wife leaves one of the parties, infidelity, and adultery, then domestic violence and ending in divorce. The dataset was taken from the West Java BPS website, and the author split the data, namely 80% training data and 20% testing data, to avoid overfitting. Research results on the classification of divorce factors during the pandemic using the C4.5 algorithm based on particle swarm optimization are an accuracy value of 87.50% and an AUC (Area Under Curve) value of 0.807, which is included in the good classification category.

Influence of Serum Apelin and CD40L Expression Levels on Adverse Cardiovascular Events after PCI

Objective: This study aimed to investigate the effects of serum levels of apelin and CD40L on major adverse cardiovascular events (MACEs) after percutaneous coronary intervention (PCI). Methods: A case–control study was conducted to select patients undergoing PCI in our hospital from June 2020 to June 2022. Patients were divided into the occurrence group and the non-occurrence group according to whether MACEs occurred during the 12-month follow-up after surgery. Enzyme-linked immunosorbent assay was used to detect the expression levels of serum apelin and CD40L in the two groups, and the correlation between the expression of apelin and CD40L and prognosis was analyzed. Logistic regression analysis was performed on the indicators with differences to analyze the influencing factors of the prognosis of PCI. Results: Compared with the non-occurrence group, the occurrence group had a significantly lower level of apelin and a significantly higher level of CD40L (p &lt; 0.001). Apelin was negatively correlated with the occurrence of MACEs after PCI (r = –0.583, p &lt; 0.001), and CD40L was positively correlated with the occurrence of MACEs after PCI (r = 0.569, p &lt; 0.001). Logistic regression analysis showed that apelin was a protective factor for MACEs after PCI (odds ratio (OR) = 0.248, p &lt; 0.001); CD40L, age, hypertension, and the number of diseased vessels were risk factors for MACEs after PCI (OR = 8.684, 0.018, 0.003, 0.020, p &lt; 0.05). The area under curve (AUC) of apelin combined with CD40L was large, and the predictive value was higher than that of apelin and CD40L alone (AUC values were 0.956, 0.857, 0.905, p &lt; 0.001; p &lt; 0.001; p &lt; 0.001). Conclusions: This study showed that the levels of apelin and CD40L were correlated with MACEs after PCI. Clinicians should pay close attention to the levels of apelin and CD40L in patients after PCI and be alert to the occurrence of MACEs.

A Lambda Layer-Based Convolutional Sequence Embedding Model for Click-Through Rate Prediction

In the era of intelligent economy, the click-through rate (CTR) prediction system can evaluate massive service information based on user historical information, and screen out the products that are most likely to be favored by users, thus realizing customized push of information and achieve the ultimate goal of improving economic benefits. Sequence modeling is one of the main research directions of CTR prediction models based on deep learning. The user's general interest hidden in the entire click history and the short-term interest hidden in the recent click behaviors have different influences on the CTR prediction results, which are highly important. In terms of capturing the user's general interest, existing models paid more attention to the relationships between item embedding vectors (point-level), while ignoring the relationships between elements in item embedding vectors (union-level). The Lambda layer-based Convolutional Sequence Embedding (LCSE) model proposed in this paper uses the Lambda layer to capture features from click history through weight distribution, and uses horizontal and vertical filters on this basis to learn the user's general preferences from union-level and point-level. In addition, we also incorporate the user's short-term preferences captured by the embedding-based convolutional model to further improve the prediction results. The AUC (Area Under Curve) values of the LCSE model on the datasets Electronic, Movie &amp; TV and MovieLens are 0.870 7, 0.903 6 and 0.946 7, improving 0.45%, 0.36% and 0.07% over the Caser model, proving the effectiveness of our proposed model.

Exosomal microRNAs associated with tuberculosis among people living with human immunodeficiency virus

ObjectiveTo investigate the diagnostic value of selected exosomal miRNAs for Tuberculosis (TB) among people living with human immunodeficiency virus (PLHIV). MethodsA total of 43 adult HIV patients, including 20 diagnosed with TB and 23 controls, were enrolled. The levels of six exosomal miRNAs (miR-20a, miR-20b, miR-26a, miR-106a, miR-191, and miR-486) were measured using qRT-PCR. ResultsThe levels of these six exosomal miRNAs (miR-20a, miR-20b, miR-26a, miR-106a, miR-191, and miR-486) were significantly higher in the plasma of TB patients compared to controls among PLHIV. The Receiver Operating Characteristic (ROC) curve of these six miRNAs showed a fair performance in distinguishing TB patients from controls, with Area Under Curve (AUC) values of 0.78 (95 %CI 0.63–0.93), 0.81 (95 %CI 0.67–0.95), 0.77 (95 %CI 0.61–0.93), 0.84 (95 %CI 0.70–0.98), 0.82 (95 %CI 0.68–0.95) and 0.79 (95 %CI 0.65–0.93), respectively. These miRNAs showed higher AUC values for extrapulmonary tuberculosis compared to pulmonary tuberculosis. An analysis of subgroups was performed based on CD4 + T cell count (＜200 and ≥ 200 cells·µL−1). In the high CD4 count group, all these six exosomal miRNAs appeared to have higher AUC values compared to the low CD4 count group. ConclusionsThese six exosomal miRNAs could serve as potential biomarkers for diagnosing TB among PLHIV.

A novel nomogram and survival analysis for different lymph node status in breast cancer based on the SEER database

IntroductionThe axillary lymph node status (ALNS) and internal mammary lymph nodes (IMLN) expression associated with breast cancer are closely linked to prognosis. This study aimed to establish a nomogram to predict survival at 3, 5, and 10 years in patients with various lymph node statuses.MethodsWe obtained data from patients with breast cancer between 2004 and 2015 from the Surveillance, Epidemiology, and End Results (SEER database). Chi-square analysis was performed to test for differences in the pathological characteristics of the groups, and Kaplan–Meier analysis and the log-rank test were used to plot and compare the correlation between overall survival (OS) and breast cancer specific survival (BCSS). The log-rank test was used for the univariate analysis, and statistically significant characteristics were included in the multivariate and Cox regression analyses. Finally, Independent factor identification was included in constructing the nomogram using R studio 4.2.0; area under curve (AUC) values were calculated, and receiver operating characteristic (ROC) curve, calibration, and decision curve analysis (DCA) curves were plotted for evaluation.ResultsA total of 279,078 patients were enrolled and analysed, demonstrating that the isolated tumour cells (ITC) group had clinicopathological characteristics similar to those of micrometastases (Mic). Multivariate analysis was performed to identify each subgroup's independent risk factors and construct a nomogram. The AUC values were 74.7 (95% CI 73.6–75.8), 72.8 (95% CI 71.9–73.8), and 71.2 (95% CI 70.2–72.2) for 3-, 5-, and 10-year OS, respectively, and 82.2 (95% CI 80.9–83.6), 80.1 (95% CI 79.0–81.2), and 75.5 (95% CI 74.3–76.8) for BCSS in overall breast cancer cases, respectively. AUC values for 3-, 5-, and 10-year OS in the ITC group were 64.8 (95% CI 56.5–73.2), 67.7 (95% CI 62.0–73.4), and 65.4 (95% CI 60.0–70.7), respectively. For those in the Mic group, AUC values for 3-, 5-, and 10-year OS were 72.9 (95% CI 70.7–75.1), 72.4 (95% CI 70.6–74.1), and 71.3 (95% CI 69.6–73.1), respectively, and AUC values for BCSS were 77.8 (95% CI 74.9–80.7), 75.7 (95% CI 73.5–77.9), and 70.3 (95% CI 68.0–72.6), respectively. In the IMLN group, AUC values for 3-, 5-, and 10-year OS were 75.2 (95% CI 71.7–78.7), 73.4 (95% CI 70.0–76.8), and 74.0 (95% CI 69.6–78.5), respectively, and AUC values for BCSS were 76.6 (95% CI 73.0–80.3), 74.1 (95% CI 70.5–77.7), and 74.7 (95% CI 69.8–79.5), respectively. The ROC, calibration, and DCA curves verified that the nomogram had better predictability and benefits.ConclusionThis study is the first to investigate the predictive value of different axillary lymph node statuses and internal mammary lymph node metastases in breast cancer, providing clinicians with additional aid in treatment decisions.

A machine learning-based model analysis for serum markers of liver fibrosis in chronic hepatitis B patients

Early assessment and accurate staging of liver fibrosis may be of great help for clinical diagnosis and treatment in patients with chronic hepatitis B (CHB). We aimed to identify serum markers and construct a machine learning (ML) model to reliably predict the stage of fibrosis in CHB patients. The clinical data of 618 CHB patients between February 2017 and September 2021 from Zhejiang Provincial People's Hospital were retrospectively analyzed, and these data as a training cohort to build the model. Six ML models were constructed based on logistic regression, support vector machine, Bayes, K-nearest neighbor, decision tree (DT) and random forest by using the maximum relevance minimum redundancy (mRMR) and gradient boosting decision tree (GBDT) dimensionality reduction selected features on the training cohort. Then, the resampling method was used to select the optimal ML model. In addition, a total of 571 patients from another hospital were used as an external validation cohort to verify the performance of the model. The DT model constructed based on five serological biomarkers included HBV-DNA, platelet, thrombin time, international normalized ratio and albumin, with the area under curve (AUC) values of the DT model for assessment of liver fibrosis stages (F0-1, F2, F3 and F4) in the training cohort were 0.898, 0.891, 0.907 and 0.944, respectively. The AUC values of the DT model for assessment of liver fibrosis stages (F0-1, F2, F3 and F4) in the external validation cohort were 0.906, 0.876, 0.931 and 0.933, respectively. The simulated risk classification based on the cutoff value showed that the classification performance of the DT model in distinguishing hepatic fibrosis stages can be accurately matched with pathological diagnosis results. ML model of five serum markers allows for accurate diagnosis of hepatic fibrosis stages, and beneficial for the clinical monitoring and treatment of CHB patients.

Value of 24h ambulatory ECG for prediction of sufficient number of premature ventricular complexes during catheter ablation

Abstract Introduction Successful catheter ablation (CA) of premature ventricular complexes (PVCs) depends on identification of the site of origin (SOO), but no parameter suggested for prediction of success from CA have been sufficiently validated in previous studies. Activation mapping and pace-mapping are invasive methods for the identification of SOO, and the burden of PVCs is critical for activation mapping to identify SOO. The number of PVCs in 24h ambulatory ECGs is claimed to predict occurrence of PVCs during the procedure, and hence procedural success, but the predictive value remains unknown. Purpose The primary objective of this study was to quantify the predictive value of pre-procedural of 24h ambulatory ECG with regard to sufficient number of PVCs during the CA procedure. Methods Patients admitted for CA of PVCs at our hospital from 2011-2020 were included. Data were retrospectivity collected from electronic health records. Clinical characteristics, results from clinical pre-procedural examinations, including 24h ambulatory ECG, exercise testing and echocardiogram, as well as acute outcomes were recorded. Results A total of 332 patients were included, CA was performed in 285 (86 %) patients, while 47 (14 %) patients had insufficient number of PVCs to allow adequate identification of SOO. Neither clinical characteristics nor results of cardiac imaging separated patients with sufficient and insufficient number of PVCs for CA, respectively. Patients with sufficient number of PVCs had nominally more PVCs in the preceding 24h ambulatory ECG, although the difference was not statistically significant (16007 (6509-26205) vs. 8332 (3066-20974), p = 0.055). A receiver operating characteristics (ROC) curve analysis of sufficient number of PVCs during CA and number of PVCs in the preceding 24h ambulatory ECG had an area under curve (AUC) of 0.610 (0.95% CI 0.498-0.722, p = 0.055). The commonly clinically used cut-off of &amp;gt;10 000 PVCs/24h had a positive predictive value of 67 % and a negative predictive value of 57 % for sufficient number of PVCs during CA. The median period from the 24h ambulatory ECGs to CA was 183 days (6 months). The positive and negative predicative values for &amp;gt;10 000 PVCs in 24h ambulatory ECG performed &amp;gt;6 months prior to the procedure were 67 % and 71 %, respectively, compared to 53 % and 61 % for 24h ECGs performed &amp;lt;6 months prior to the procedure. The ROC curve for sufficient number of PVCs during the procedure and number of PVCs in a 24h ambulatory ECG performed &amp;gt;6 months and &amp;lt;6 months prior to the procedure did not differ, with had AUC values of 0.665 and 0.584, respectively (p = 0.4). Conclusion In this retrospective analysis of a large cohort of patients admitted for CA of PVCs, number of PVCs in 24h ambulatory ECGs recorded prior to the procedure had a low predictive value for presence of PVCs during CA. Others parameters need to be evaluated to improve prediction.

“Percentage” and “size” of residual viable tumor in lymph node, the performance in estimating pathologic response of lymph node in non-small cell lung cancer treated with neoadjuvant chemoimmunotherapy

There is no universally accepted method for evaluating lymph node metastasis (LNM) in non-small cell lung cancer (NSCLC) after neoadjuvant chemoimmunotherapy. Different protocols recommend evaluating the percentage of residual viable tumor (RVT%) and metastatic tumor size (MTS). Our aim was to determine the prognostic significance of RVT% and MTS, and identify the more effective parameter for pathological evaluating LNM. Two independent cohorts were collected (derivation, n = 84; external validation, n = 42). All patients exhibited metastatic cancer or treatment response in lymph nodes post-surgery. In the derivation cohort, we assessed the mean and largest values of MTS and RVT% in LNM, estimating their optimal cutoffs for event-free survival (EFS) using maximally selected rank statistics. Validation was subsequently conducted in the external validation cohort. The quality of prognostic factors was evaluated using the Area Under Curve (AUC). A positive association was identified between RVT% and MTS, but an absolute association could not be conclusively established. In the derivation cohort, neither the largest MTS (cutoff = 6 mm, p = 0.28), largest RVT% (cutoff = 75%, p = 0.23), nor mean RVT% (cutoff = 55%, p = 0.06) were associated with EFS. However, mean MTS (cutoff = 4.5 mm) in lymph nodes was statistically associated with EFS (p = 0.018), validated by the external cohort (p = 0.017). The prognostic value of MTS exceeded that of ypN staging in both cohorts, as evidenced by higher AUC values. The mean value of MTS can effectively serve as a parameter for the pathological evaluation of lymph nodes, with a threshold of 4.5 mm, closely linked to EFS. Its prognostic value outperforms that of ypN staging.

Development of a Novel Lipid Metabolism-related Gene Prognostic Signature for Patients with Colorectal Cancer.

The purpose of this study was to explore the expression profiles of lipid metabolism-related genes in patients with Colorectal Cancer (CRC). The lipid metabolism statuses of CRC patients from The Cancer Genome Atlas (TCGA) were analyzed. Risk characteristics were constructed by univariate Cox regression and minimum Absolute contraction and Selection Operator (LASSO) Cox regression. A histogram was constructed based on factors such as age, sex, TNM stage, T stage, N stage, and risk score to provide a visual tool for clinicians to predict the probability of 1-year, 3-year, and 5-year OS for CRC patients. By determining Area Under Curve (AUC) values, the time-dependent Receiver Operating characteristic Curve (ROC) was used to evaluate the efficiency of our model in predicting prognosis. A novel risk signal based on lipid metabolism-related genes was constructed to predict the survival of CRC patients. Risk characteristics were shown to be an independent prognostic factor in CRC patients (p <0.001). There were significant differences in the abundance and immune characteristics of tumor-filtering immune cells between high-risk and low-risk groups. The nomogram had a high potential for clinical application and the ROC AUC value was 0.827. Moreover, ROC analysis demonstrated that the nomogram model was more accurate to predict the survival of CRC patients than age, gender, stage and risk score. In this study, we demonstrated a lipid metabolism-related genes prognosis biomarker associated with the tumor immune micro-environment in patients with CRC.

Clinical characteristics of liver transplant recipients with COVID-19 and analysis of risk factors for the severe disease.

Liver transplant (LT) recipients were at a high risk of infection during the coronavirus disease 2019 (COVID-19) pandemic. Our purpose was to compare the clinical characteristics of severe and non-severe groups of LT recipients with COVID-19, and to analyze their risk factors for severe disease. 79 LT recipients with COVID-19 were divided into a non-severe group (n = 60) and a severe group (n = 19), and differences in clinical characteristics, laboratory tests, and chest computed tomography (CT) performance were analyzed. Logistic regression was used to identify risk factors with severe COVID-19. Receiver operating characteristic (ROC) curves were plotted and the area under curve (AUC) values were calculated to assess the predictive value for severe COVID-19. Age was statistically different (p < 0.001) between the two groups. The difference in neutrophil-to-lymphocyte ratio (NLR), serum creatinine (Scr), D-dimer, urea, C-reactive protein (CRP), lactate dehydrogenase (LDH), and the number of lung segments involved in inflammation between the two groups were statistically significant (p < 0.05). The results revealed that age (OR = 1.255, 95% CI 1.079-1.460), NLR (OR = 1.172, 95% CI 1.019-1.348), and Scr (OR = 1.041, 95% CI 1.016-1.066) were independent risk factors for severe COVID-19. The ROC results showed that high values for age, NLR and Scr predicted severe COVID-19, with AUC values of 0.775, 0.841 and 0.820, respectively, and 0.925 for the three factors combined. Advanced age, and elevated NLR and Scr are independent risk factors for severe COVID-19 in LT recipients.

Enhancing vulnerability assessment through spatially explicit modeling of mountain social-ecological systems exposed to multiple environmental hazards

The evaluation of the vulnerability of coupled socio-ecological systems is critical for addressing and preventing the adverse impacts of various environmental hazards and devising strategies for climate change adaptation. The initial step in vulnerability assessment involves exposure assessment, which entails quantifying and mapping the risks posed by multiple environmental hazards, thereby offering valuable insights for the implementation of vulnerability assessment methodologies. Consequently, this study sought to model the exposure of coupled social-ecological systems in mountainous regions to various environmental hazards. By a set of socio-economic, climatic, geospatial, hydrological, and demographic data, as well as satellite imagery, and examining 11 hazards, including droughts, pests, dust storms, winds, extreme temperatures, evapotranspiration, landslides, floods, wildfires, and social vulnerability, this research employed machine learning (ML) techniques and the fuzzy analytical hierarchy process (FAHP). Expert opinions were utilized to guide hazard weighting and calculate the exposure index (EI). Through the precise spatial mapping of EI variations across the socio-ecological systems in mountainous areas, this investigation provides insights into vulnerability to multiple environmental hazards, thereby laying the groundwork for future endeavors in supporting national-level vulnerability assessments aimed at fostering sustainable environments. The findings reveal that social vulnerability and pests receive the highest weighting, while floods and landslides are ranked lower. All hazards demonstrate significant correlations with the EI, with droughts exhibiting the strongest correlation (r > 0.81). Spatial analysis indicates a north-south gradient in forest exposure, with southern regions showing higher exposure hotspots (EI 29.08) compared to northern areas (EI 10.60). Validation based on Area Under Curve (AUC) and Consistency Rate (CR) in FAHP demonstrates robustness, with AUC values exceeding 0.78 and CR values below 0.1. Considering the anticipated intensification of hazards, management strategies should prioritize reducing social vulnerability, restore degraded areas using drought-resistant species, combat pests, and mitigate desertification. By integrating multidisciplinary data and expert opinions, this research contributes to informed decision-making regarding sustainable forest management and climate resilience in mountain ecosystems.

Noninvasive PCA3 detection for screening prostate diseases using a graphene oxide-based method

Prostate Cancer Associated 3 (PCA3) as a promising biomarker to replace serum prostate-specific antigen (PSA) for differential diagnosis of benign and malignant prostate diseases. However, the detection of PCA3 in urine requires the collection of digital rectal examination (DRE) urine, which limits the application of PCA3 in screening prostate diseases. In the present study, a new reverse transcription quantitative real-time polymerase chain reaction (qRT-PCR) method was developed for PCA3 detection in non-DRE urine. The method uses the nanomaterial Graphene oxide (GO) to effectively prevent nonspecific amplification in qRT-PCR. GO-based qRT-PCR produced no nonspecific amplification. Receiver operating characteristic curve (ROC) results showed that the area under curve (AUC) of PCA3 to discriminate normal control subjects from prostate cancer (PCa) was 0.869 (95% confidence interval [CI], 0.689–1.000), and using 2.711 as the cutoff value of PCA3, the sensitivity was 81.8% and the specificity was as high as 88.9%. For prostatitis (PTIS) and PCa, the AUC value of PCA3 to distinguish them was 0.636 (95% CI, 0.396–0.877), with 81.8% sensitivity and 58.3% specificity. Compared with the 8.3% specificity of PAS, the specificity of PCA3 was improved by 50%, which suggested that PCA3, in non-DRE urine, has a significant potential application in the differential diagnosis of prostate diseases.

Screening of lung cancer serum biomarkers based on Boruta-shap and RFC-RFECV algorithms

ObjectiveThis study aimed to identify a set of serum miRNAs as potential biomarkers for lung cancer diagnosis using algorithmic approaches. MethodsSerum miRNA expression data from lung cancer patients and non-tumor controls were obtained. The top six miRNAs were selected using Boruta-shap and RFC-RFECV algorithms. A Gaussian Naive Bayes (NB) classifier was trained and evaluated using cross-validation, ROC curve analysis, and evaluation metrics. ResultsSix miRNAs (hsa-miRNA-144, hsa-miRNA-107, hsa-miRNA-484, hsa-miRNA-103, hsa-miRNA-26b, and hsa-miRNA-641) were identified as feature genes. The NB classifier achieved an area under curve (AUC) of 0.8966 and a mean AUC of 0.88 in cross-validation. Accuracy, recall, and F1 scores exhibited promising results, with an accuracy of 82%. In the validation set, the AUC values for the NB and SVC classifiers were 0.9345 and 0.9423, respectively, with a mean AUC of 0.95 in cross-validation. The classifiers demonstrated an accuracy of 89% in diagnosing lung cancer. ConclusionThis study identified a panel of six serum miRNAs with potential as non-invasive biomarkers for lung cancer diagnosis. These miRNAs show promise in providing sensitive and specific tools for detecting lung cancer. SignificanceLung cancer is one of the top cancers worldwide, threatening the health and lives of tens of thousands of people. miRNA is a biomarker, which can be used as a potential clinical tool for diagnosis and prognosis of cancer patients. Therefore, the use of multiple miRNAs to construct diagnostic models may be one of the future methods of accurate diagnosis of lung cancer. In this study, we used the Boruta-shap and RFC-RFECV algorithms to automatically identify and extract characteristic miRNAs highly associated with lung cancer, thereby establishing an accurate classifier for the diagnosis of lung cancer with characteristic miRNAs.

Deep learning-assisted diagnosis of benign and malignant parotid tumors based on ultrasound: a retrospective study

BackgroundTo develop a deep learning(DL) model utilizing ultrasound images, and evaluate its efficacy in distinguishing between benign and malignant parotid tumors (PTs), as well as its practicality in assisting clinicians with accurate diagnosis.MethodsA total of 2211 ultrasound images of 980 pathologically confirmed PTs (Training set: n = 721; Validation set: n = 82; Internal-test set: n = 89; External-test set: n = 88) from 907 patients were retrospectively included in this study. The optimal model was selected and the diagnostic performance evaluation is conducted by utilizing the area under curve (AUC) of the receiver-operating characteristic(ROC) based on five different DL networks constructed at varying depths. Furthermore, a comparison of different seniority radiologists was made in the presence of the optimal auxiliary diagnosis model. Additionally, the diagnostic confusion matrix of the optimal model was calculated, and an analysis and summary of misjudged cases’ characteristics were conducted.ResultsThe Resnet18 demonstrated superior diagnostic performance, with an AUC value of 0.947, accuracy of 88.5%, sensitivity of 78.2%, and specificity of 92.7% in internal-test set, and with an AUC value of 0.925, accuracy of 89.8%, sensitivity of 83.3%, and specificity of 90.6% in external-test set. The PTs were subjectively assessed twice by six radiologists, both with and without the assisted of the model. With the assisted of the model, both junior and senior radiologists demonstrated enhanced diagnostic performance. In the internal-test set, there was an increase in AUC values by 0.062 and 0.082 for junior radiologists respectively, while senior radiologists experienced an improvement of 0.066 and 0.106 in their respective AUC values.ConclusionsThe DL model based on ultrasound images demonstrates exceptional capability in distinguishing between benign and malignant PTs, thereby assisting radiologists of varying expertise levels to achieve heightened diagnostic performance, and serve as a noninvasive imaging adjunct diagnostic method for clinical purposes.

Continuous Cardiac Magnetic Resonance Imaging after Coronary Revascularization for Left Ventricular Dysfunction

Objective: To determine the contribution of serial cardiac magnetic resonance imaging (MRI) following coronary revascularization (CR) to the clinical management of patients with left ventricular insufficiency. Methods: The study objects comprised the clinical data of 145 patients with CR undergoing CR surgery for left ventricular insufficiency in our hospital from January 2021 to January 2023. The patients were divided into the case (n = 35, left ventricular ejection fraction (LVEF) &lt;50%) and control (n = 110, LVEF ≥50%) groups based on the LVEF recorded in the medical record system 6 months after surgery. Preoperative LVEF left ventricular end-diastolic volume (LVEDV), left ventricular end-systolic volume (LVESV), left ventricular end-diastolic volume index (LVEDVI), left ventricular end-systolic volume index (LVESVI), cardiac index (CI), and other cardiac magnetic resonance detection parameters were compared. Logistic regression analysis was performed to analyze the prognostic factors of patients undergoing CR after CR surgery for left ventricular insufficiency. The receiver operating characteristic curve was drawn, the sensitivity, specificity, and area under curve (AUC) were calculated, and the best prediction threshold was determined. The prognostic value of cardiac MRI in CR surgery for left ventricular dysfunction was observed. Results: Cardiac MRI revealed that the case group had higher LVEDV, LVESV, LVEDVI, LVESVI, and CI than the control group. However, the LVEF index was lower than that in the control group (p &lt; 0.05). Logistic regression analysis was conducted for indicators with differences, and the results indicate LVEF as a protective factor for the postoperative efficacy of the patients, with an odds ratio (OR) &lt;1. LVEDV, LVESV, LVEDVI, LVESVI, and CI were all risk factors for the postoperative efficacy of the patients, with an OR &gt;1. The AUC values of LVEF, LVEDV, LVESV, LVEDVI, LVESVI, and CI were 0.698, 0.674, 0.654, 0.700, 0.572, and 0.812, respectively. The optimal threshold values were 53.57%, 112.33 and 68.5 mL, and 205.51, 163.99, and 2.14 L/m2, and their corresponding sensitivities reached 0.618, 0.514, 0.654, 0.800, 0.371, and 0.829 for each index. The specificities were 0.800, 0.836, 0.771, 0.609, 0.836, and 0.645, which indicate that LVEF, LVEDV, LVESV, LVEDVI, LVESVI, and CI had a certain degree of predictive value for postoperative cardiac function recovery. Conclusion: LVEDV, LVESV, LVEDVI, LVESVI, CI, and LVEF are all factors affecting the clinical efficacy in patients undergoing CR after left ventricular insufficiency. In addition, cardiac MRI can effectively detect the above factors and effectively predict the postoperative efficacy among patients.

The relationship between heavy metals and metabolic syndrome using machine learning.

Exposure to high levels of heavy metals has been widely recognized as an important risk factor for metabolic syndrome (MetS). The main purpose of this study is to assess the associations between the level of heavy metal exposure and Mets using machine learning (ML) method. The data used in this study are from the national health and nutrition examination survey 2003-2018. According to the demographic information and heavy metal exposure level of participants, a total of 22 variables were included. Lasso was used to screen out the key variables, and 9 commonly used ML models were selected to establish the associations with the 5-fold cross validation method. Finally, we choose the SHapley Additive exPlanations (SHAP) method to explain the prediction results of Adaboost model. 11,667 eligible individuals were randomly divided into two groups to train and verify the prediction model. Through lasso, characteristic variables were selected from 24 variables as predictors. The AUC (area under curve) of the models selected in this study were all greater than 0.7, and AdaBoost was the best model. The AUC value of AdaBoost was 0.807, the accuracy was 0.720, and the sensitivity was 0.792. It is noteworthy that higher levels of cadmium, body mass index, cesium, being female, and increasing age were associated with an increased probability of MetS. Conversely, lower levels of cobalt and molybdenum were linked to a decrease in the estimated probability of MetS. Our study highlights the AdaBoost model proved to be highly effective, precise, and resilient in detecting a correlation between exposure to heavy metals and MetS. Through the use of interpretable methods, we identified cadmium, molybdenum, cobalt, cesium, uranium, and barium as prominent contributors within the predictive model.

A multi-instance tumor subtype classification method for small PET datasets using RA-DL attention module guided deep feature extraction with radiomics features

Background:Positron emission tomography (PET) is extensively employed for diagnosing and staging various tumors, including liver cancer, lung cancer, and lymphoma. Accurate subtype classification of tumors plays a crucial role in formulating effective treatment plans for patients. Notably, lymphoma comprises subtypes like diffuse large B-cell lymphoma and Hodgkin’s lymphoma, while lung cancer encompasses adenocarcinoma, small cell carcinoma, and squamous cell carcinoma. Similarly, liver cancer consists of subtypes such as cholangiocarcinoma and hepatocellular carcinoma. Consequently, the subtype classification of tumors based on PET images holds immense clinical significance. However, in clinical practice, the number of cases available for each subtype is often limited and imbalanced. Therefore, the primary challenge lies in achieving precise subtype classification using a small dataset. Method:This paper presents a novel approach for tumor subtype classification in small datasets using RA-DL (Radiomics-DeepLearning) attention. To address the limited sample size, Support Vector Machines (SVM) is employed as the classifier for tumor subtypes instead of deep learning methods. Emphasizing the importance of texture information in tumor subtype recognition, radiomics features are extracted from the tumor regions during the feature extraction stage. These features are compressed using an autoencoder to reduce redundancy. In addition to radiomics features, deep features are also extracted from the tumors to leverage the feature extraction capabilities of deep learning. In contrast to existing methods, our proposed approach utilizes the RA-DL-Attention mechanism to guide the deep network in extracting complementary deep features that enhance the expressive capacity of the final features while minimizing redundancy. To address the challenges of limited and imbalanced data, our method avoids using classification labels during deep feature extraction and instead incorporates 2D Region of Interest (ROI) segmentation and image reconstruction as auxiliary tasks. Subsequently, all lesion features of a single patient are aggregated into a feature vector using a multi-instance aggregation layer. Result:Validation experiments were conducted on three PET datasets, specifically the liver cancer dataset, lung cancer dataset, and lymphoma dataset. In the context of lung cancer, our proposed method achieved impressive performance with Area Under Curve (AUC) values of 0.82, 0.84, and 0.83 for the three-classification task. For the binary classification task of lymphoma, our method demonstrated notable results with AUC values of 0.95 and 0.75. Moreover, in the binary classification task of liver tumor, our method exhibited promising performance with AUC values of 0.84 and 0.86. Conclusion:The experimental results clearly indicate that our proposed method outperforms alternative approaches significantly. Through the extraction of complementary radiomics features and deep features, our method achieves a substantial improvement in tumor subtype classification performance using small PET datasets.