Shock index to predict outcomes in patients with trauma following traffic collisions: a retrospective cohort study.

PURPOSE: Taiwan, which has a rate of high vehicle ownership, faces significant challenges in managing trauma caused by traffic collisions. In Taiwan, traffic collisions contribute significantly to morbidity and mortality, with a high incidence of severe bleeding trauma. The shock index (SI) and the modified shock index (MSI) have been proposed as early indicators of hemodynamic instability. In this study, we aimed to assess the efficacy of SI and MSI in predicting adverse outcomes in patients with trauma following traffic collisions. METHODS: This retrospective cohort study was conducted at Chi Mei Hospital from January 2015 to December 2020. The comprehensive analysis included 662 patients, with data collected on vital signs and outcomes such as mortality, blood transfusion, emergent surgical intervention (ESI), transarterial embolization (TAE), and intensive care unit (ICU) admission. Optimal cutoff points for SI and MSI were identified by calculating the Youden index. Logistic regression analysis was used to assess outcomes, adjusting for demographic and injury severity variables. RESULTS: An SI threshold of 1.11 was associated with an increased risk of mortality, while an SI of 0.84 predicted the need for blood transfusion in the context of traffic collisions. Both SI and MSI demonstrated high predictive power for mortality and blood transfusion, with acceptable accuracy for TAE, ESI, and ICU admission. Logistic regression analyses confirmed the independence of SI and MSI as risk factors for adverse outcomes, thus, providing valuable insights into their clinical utility. CONCLUSIONS: SI and MSI are valuable tools for predicting mortality and blood transfusion needs in patients with trauma due to traffic collisions. These findings advance the quality of care for patients with trauma during their transition from the emergency room to the ICU, facilitating prompt and reliable decision-making processes and improving the care of patients with trauma.

Medical diagnostic accuracy measures: an innovative approach based on the area under predictive values curves.

Positive and negative estimates are commonly used by clinicians to evaluate the likelihood of a disease stage being present based on test results. The predicted values are dependent on the prevalence of the underlying illness. However, for certain diseases or clinical conditions, the prevalence is unknown or different from one region to another or from one population to another, leading to an erroneous diagnosis. This article introduces innovative post-test diagnostic precision measures for continuous tests or biomarkers based on the combined areas under the predictive value curves for all possible prevalence values. The proposed measures do not vary as a function of the prevalence of the disease. They can be used to compare different diagnostic tests and/or biomarkers' abilities for rule-in, rule-out, and overall accuracy based on the combined areas under the predictive value curves. The relationship of the proposed measures to other diagnostic accuracy measures is discussed. We illustrate the proposed measures numerically and use a real data example on breast cancer.

Methods of determining optimal cut-point of diagnostic biomarkers with application of clinical data in ROC analysis: an update review.

INTRODUCTION: An important application of ROC analysis is the determination of the optimal cut-point for biomarkers in diagnostic studies. This comprehensive review provides a framework of cut-point election for biomarkers in diagnostic medicine. METHODS: Several methods were proposed for the selection of optional cut-points. The validity and precision of the proposed methods were discussed and the clinical application of the methods was illustrated with a practical example of clinical diagnostic data of C-reactive protein (CRP), erythrocyte sedimentation rate (ESR) and malondialdehyde (MDA) for prediction of inflammatory bowel disease (IBD) patients using the NCSS software. RESULTS: Our results in the clinical data suggested that for CRP and MDA, the calculated cut-points of the Youden index, Euclidean index, Product and Union index methods were consistent in predicting IBD patients, while for ESR, only the Euclidean and Product methods yielded similar estimates. However, the diagnostic odds ratio (DOR) method provided more extreme values for the optimal cut-point for all biomarkers analyzed. CONCLUSION: Overall, the four methods including the Youden index, Euclidean index, Product, and IU can produce quite similar optimal cut-points for binormal pairs with the same variance. The cut-point determined with the Youden index may not agree with the other three methods in the case of skewed distributions while DOR does not produce valid informative cut-points. Therefore, more extensive Monte Carlo simulation studies are needed to investigate the conditions of test result distributions that may lead to inconsistent findings in clinical diagnostics.

Combining multiple biomarkers linearly to minimize the Euclidean distance of the closest point on the receiver operating characteristic surface to the perfection corner in trichotomous settings.

The performance of individual biomarkers in discriminating between two groups, typically the healthy and the diseased, may be limited. Thus, there is interest in developing statistical methodologies for biomarker combinations with the aim of improving upon the individual discriminatory performance. There is extensive literature referring to biomarker combinations under the two-class setting. However, the corresponding literature under a three-class setting is limited. In our study, we provide parametric and nonparametric methods that allow investigators to optimally combine biomarkers that seek to discriminate between three classes by minimizing the Euclidean distance from the receiver operating characteristic surface to the perfection corner. Using this Euclidean distance as the objective function allows for estimation of the optimal combination coefficients along with the optimal cutoff values for the combined score. An advantage of the proposed methods is that they can accommodate biomarker data from all three groups simultaneously, as opposed to a pairwise analysis such as the one implied by the three-class Youden index. We illustrate that the derived true classification rates exhibit narrower confidence intervals than those derived from the Youden-based approach under a parametric, flexible parametric, and nonparametric kernel-based framework. We evaluate our approaches through extensive simulations and apply them to real data sets that refer to liver cancer patients.

Performance of diagnostic tests based on continuous bivariate markers.

In medical diagnostic research, it is customary to collect multiple continuous biomarker measures to improve the accuracy of diagnostic tests. A prevalent practice is to combine the measurements of these biomarkers into one single composite score. However, incorporating those biomarker measurements into a single score depends on the combination of methods and may lose vital information needed to make an effective and accurate decision. Furthermore, a diagnostic cut-off is required for such a combined score, and it is difficult to interpret in actual clinical practice. The paper extends the classical biomarkers' accuracy and predictive values from univariate to bivariate markers. Also, we will develop a novel pseudo-measures system to maximize the vital information from multiple biomarkers. We specified these pseudo-and-or classifiers for the true positive rate, true negative rate, false-positive rate, and false-negative rate. We used them to redefine classical measures such as the Youden index, diagnostics odds ratio, likelihood ratios, and predictive values. We provide optimal cut-off point selection based on the modified Youden index with numerical illustrations and real data analysis for this paper's newly developed pseudo measures.

Using the Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST) to Predict Gambling Disorder Among U.S. College Students.

The Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST) screening tool has not previously been used to evaluate risk for gambling disorder (GD). We aimed to assess the level at which each specific substance involvement score (SSIS), measured by ASSIST, most optimally predicted GD among U.S. college students. Data were analyzed for 141,769 students from the National College Health Assessment (fall 2019-spring 2021) utilizing multivariable logistic regression models. Sensitivities and specificities were utilized to find optimal cutoffs that best identified those with GD, overall and by biological sex and age group. Lower threshold of substance risk related to prescription opioids, cocaine, and hallucinogens (all with SSIS cutoffs of 4) predicts gambling disorder compared to sedatives (SSIS cutoff of 19). Younger students had lower thresholds of substance risk predicting GD than older students for heroin, but for all other substance classifications students 25 years and older had lower thresholds of SSIS predicting GD than students 18-24 years old. This study aids in the understanding that substance use behavior may put students at risk for other addictive behaviors such as GD. This study is the first to utilize the ASSIST tool to predict GD among U.S. college students, extending its application beyond substance use disorders. The identification of optimal cutoffs for each SSIS provides a novel approach to concurrently screen for GD and substance use disorders. This unique contribution could enhance early detection and intervention strategies for GD in the college student population.

A Step-by-Step Guide to Selecting an Optimal Cut-Off Value Based on the Receiver Operating Characteristic and Youden Index in Methods Designed to Diagnose Lyme Disease.

Detection tools designed to diagnose complex diseases such as Lyme Borreliosis require an optimal cutoff point to distinguish the healthy from the diseased. The chapter will provide a practical guide to selecting an optimal cutoff mark by creating the receiver operating characteristic (ROC) in Microsoft Excel. To guide the creation of a ROC graphical plot, we will use example data from an enzyme-linked immunosorbent assay (ELISA) measuring anti-human immunoglobulin G (IgG) against whole-cell Borrelia lysates. Herein, the ROC method will demonstrate that an optical density (OD) value from ELISA with the highest Youden Index (J) is an optimal cutoff value to differentiate positive and negative IgG immune responses in human serum samples.

Benchmarking MicrobIEM - a user-friendly tool for decontamination of microbiome sequencing data.

BACKGROUND: Microbiome analysis is becoming a standard component in many scientific studies, but also requires extensive quality control of the 16S rRNA gene sequencing data prior to analysis. In particular, when investigating low-biomass microbial environments such as human skin, contaminants distort the true microbiome sample composition and need to be removed bioinformatically. We introduce MicrobIEM, a novel tool to bioinformatically remove contaminants using negative controls. RESULTS: We benchmarked MicrobIEM against five established decontamination approaches in four 16S rRNA amplicon sequencing datasets: three serially diluted mock communities (108-103 cells, 0.4-80% contamination) with even or staggered taxon compositions and a skin microbiome dataset. Results depended strongly on user-selected algorithm parameters. Overall, sample-based algorithms separated mock and contaminant sequences best in the even mock, whereas control-based algorithms performed better in the two staggered mocks, particularly in low-biomass samples (≤ 106 cells). We show that a correct decontamination benchmarking requires realistic staggered mock communities and unbiased evaluation measures such as Youden's index. In the skin dataset, the Decontam prevalence filter and MicrobIEM's ratio filter effectively reduced common contaminants while keeping skin-associated genera. CONCLUSIONS: MicrobIEM's ratio filter for decontamination performs better or as good as established bioinformatic decontamination tools. In contrast to established tools, MicrobIEM additionally provides interactive plots and supports selecting appropriate filtering parameters via a user-friendly graphical user interface. Therefore, MicrobIEM is the first quality control tool for microbiome experts without coding experience.

Optimal Cut-Off Values for Body Mass Index and Fat Mass Index Based on Age in Physically Active Males Using Receiver Operating Characteristic Curve.

This study aims to redefine obesity cut-off points for body mass index (BMI) and fat mass index (FMI) according to the different age groups of physically active males. Healthy physically active volunteers (N = 1442) aged 18-57 years (y), with a mean BMI = 22.7 ± 2.8 kg/m2, and mean FMI = 4.3 ± 1.7 kg/m2 were recruited from various fitness centers. BMI was calculated and individuals were categorized according to the Asia-Pacific BMI criterion of ≤22.9 kg/m2 and the previous WHO-guided BMI criterion of ≤24.9 kg/m2. FMI was also calculated for the study participants with a cut-off of 6.6 kg/m2. Redefining of BMI and FMI cut-off values was carried out based on different age groups categorized with a difference of 10 y and 5 y using the receiver operating characteristic (ROC) curve and Youden's index. For the entire study population, BMI redefined cut-off points for overweight and obesity were 23.7 kg/m2 and 24.5 kg/m2, respectively, while FMI redefined cut-off points for overweight and obesity were 4.6 kg/m2 and 5.7 kg/m2, respectively. With 10 y of age group difference, a constant BMI and FMI values were observed, while with 5 y of age group difference, a constant increase in the BMI cut-offs was observed as the age group increased, i.e., from 23.3 kg/m2 in 20-24 y to 26.6 kg/m2 in ≥45 y and a similar trend was seen in FMI cut-offs. To conclude, our study suggests that age-dependent BMI and FMI cut-off points may provide appropriate measurements for physically active males as the age group increases.

Estimating transformations for evaluating diagnostic tests with covariate adjustment.

Receiver operating characteristic analysis is one of the most popular approaches for evaluating and comparing the accuracy of medical diagnostic tests. Although various methodologies have been developed for estimating receiver operating characteristic curves and their associated summary indices, there is no consensus on a single framework that can provide consistent statistical inference while handling the complexities associated with medical data. Such complexities might include non-normal data, covariates that influence the diagnostic potential of a test, ordinal biomarkers or censored data due to instrument detection limits. We propose a regression model for the transformed test results which exploits the invariance of receiver operating characteristic curves to monotonic transformations and accommodates these features. Simulation studies show that the estimates based on transformation models are unbiased and yield coverage at nominal levels. The methodology is applied to a cross-sectional study of metabolic syndrome where we investigate the covariate-specific performance of weight-to-height ratio as a non-invasive diagnostic test. Software implementations for all the methods described in the article are provided in the tram add-on package to the R system for statistical computing and graphics.

Statistical inference for the two-sample problem under likelihood ratio ordering, with application to the ROC curve estimation.

The receiver operating characteristic (ROC) curve is a powerful statistical tool and has been widely applied in medical research. In the ROC curve estimation, a commonly used assumption is that larger the biomarker value, greater severity the disease. In this article, we mathematically interpret "greater severity of the disease" as "larger probability of being diseased." This in turn is equivalent to assume the likelihood ratio ordering of the biomarker between the diseased and healthy individuals. With this assumption, we first propose a Bernstein polynomial method to model the distributions of both samples; we then estimate the distributions by the maximum empirical likelihood principle. The ROC curve estimate and the associated summary statistics are obtained subsequently. Theoretically, we establish the asymptotic consistency of our estimators. Via extensive numerical studies, we compare the performance of our method with competitive methods. The application of our method is illustrated by a real-data example.

Screening of coexisting depression or anxiety disorders among Danish patients with alcohol use disorder in outpatient treatment.

BACKGROUND: Depression and anxiety are highly prevalent among patients seeking outpatient treatment for alcohol use disorders (AUD) and if depression and anxiety are addressed the prognosis is improved. Screening instruments for depression and anxiety have been validated in populations suffering from drug use disorders, but not in populations suffering from AUD. The aim of this study was to validate four self-administrated screening instruments (PHQ-9, GAD-7, Kessler-6, and SRQ) and calculate the optimal cut-off value for identifying depression and anxiety. METHODS: The study included 73 patients with self-reported depression or anxiety during AUD treatment. Each patient filled out the above-mentioned instruments and was subsequently interviewed by trained clinicians blinded to the results of the instruments with the Present State Examination to establish a diagnosis of depression or anxiety according to ICD-10. ROC curves were constructed for each instrument and the area under the curve (AUC) was calculated using patients with no depression or anxiety as reference. Youden's index was calculated to assess the optimal cut-off for each instrument. RESULTS: A total of 33 (45.2%) were diagnosed with depression or anxiety. The AUC for PHQ-9, GAD-7, Kessler-6, and SRQ were 0.767, 0.630, 0.793, and 0.698 respectively. Kessler-6, the instruments performing best based on the AUC, identified 27 (82%) of the 33 patients using a cut-off of 10 points. CONCLUSION: Kessler-6 seems to be valid and reliable in identifying patients requiring treatment for depression or anxiety among patients seeking treatment for AUD who are reporting depression or anxiety.

Discovery of a biomarker for ß-Thalassemia by HPLC-MS and improvement from Proton Transfer Reaction - Parallel Ion Parking.

ß-thalassemia is a quantitative hemoglobin (Hb) disorder resulting in reduced production of Hb A and increased levels of Hb A2. Diagnosis of ß-thalassemia can be problematic when combined with other structural Hb variants, so that the separation approaches in routine clinical centers are not sufficiently decisive to obtain accurate results. Here, we separate the intact Hb subunits by high-performance liquid chromatography, followed by top-down tandem mass spectrometry of intact subunits to distinguish Hb variants. Proton transfer reaction-parallel ion parking (PTR-PIP), in which a radical anion removes protons from multiply charged precursor ions and produces charge-reduced ions spanning a limited m/z range, was used to increase the signal-to-noise ratio of the subunits of interest. We demonstrate that the δ/ß ratio can act as a biomarker to identify ß-thalassemia in normal electrospray ionization MS1 and PTR-PIP MS1. The application of PTR-PIP significantly increases the sensitivity and specificity of the HPLC-MS method to identify δ/ß ratio as a thalassemia biomarker.

A tale of two recession-derivative indicators.

Two recession-derivative indicators (RDIs) have been used extensively as forecast objects in business cycle prediction, viz. (1) the target variable takes value 1 if there is a recession starting exactly at a specific horizon in the future, and (2) the target variable takes value 1 if there is a recession starting any time over a specified period in the future. Using daily yield spread as an illustrative predictor, we formally and quantitatively compare the two RDIs using the receiver operating characteristics analysis. Over 1962-2021 covering eight NBER recessions, we find that generally the second RDI, ceteris paribus, will make the the predictor better performing. However, the first RDI can generate better-looking and more useful predictions under certain scenarios, depending on forecast horizon, recession duration and time profile of signals. We also consider a semiannual chronology proposed by Peláez (J Macroecon 45:384-393, 2015) and find that its performance is in the middle of the other two. Our analysis suggests that the choice of a particular RDI should be dictated by the needs of forecast user in a particular decision making context.

Comparison of clinician diagnosis of COVID-19 with real time polymerase chain reaction in an adult-representative population in Sweden.

BACKGROUND: Due to the high transmissibility of SARS-CoV-2, accurate diagnosis is essential for effective infection control, but the gold standard, real-time reverse transcriptase-polymerase chain reaction (RT-PCR), is costly, slow, and test capacity has at times been insufficient. We compared the accuracy of clinician diagnosis of COVID-19 against RT-PCR in a general adult population. METHODS: COVID-19 diagnosis data by 30th September 2021 for participants in an ongoing population-based cohort study of adults in Western Sweden were retrieved from registers, based on positive RT-PCR and clinician diagnosis using recommended ICD-10 codes. We calculated accuracy measures of clinician diagnosis using RT-PCR as reference for all subjects and stratified by age, gender, BMI, and comorbidity collected pre-COVID-19. RESULTS: Of 42,621 subjects, 3,936 (9.2%) and 5705 (13.4%) had had COVID-19 identified by RT-PCR and clinician diagnosis, respectively. Sensitivity and specificity of clinician diagnosis against RT-PCR were 78% (95%CI 77-80%) and 93% (95%CI 93-93%), respectively. Positive predictive value (PPV) was 54% (95%CI 53-55%), while negative predictive value (NPV) was 98% (95%CI 98-98%) and Youden's index 71% (95%CI 70-72%). These estimates were similar between men and women, across age groups, BMI categories, and between patients with and without asthma. However, while specificity, NPV, and Youden's index were similar between patients with and without chronic obstructive pulmonary disease (COPD), sensitivity was slightly higher in patients with (84% [95%CI 74-90%]) than those without (78% [95%CI 77-79%]) COPD. CONCLUSIONS: The accuracy of clinician diagnosis for COVID-19 is adequate, regardless of gender, age, BMI, and asthma, and thus can be used for screening purposes to supplement RT-PCR.

SARS-CoV-2 Serosurveys: How Antigen, Isotype and Threshold Choices Affect the Outcome.

BACKGROUND: Evaluating the performance of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) serological assays and clearly articulating the utility of selected antigens, isotypes, and thresholds is crucial to understanding the prevalence of infection within selected communities. METHODS: This cross-sectional study, implemented in 2020, screened PCRconfirmed coronavirus disease 2019 patients (n 86), banked prepandemic and negative samples (n 96), healthcare workers and family members (n 552), and university employees (n 327) for antiSARS-CoV-2 receptor-binding domain, trimeric spike protein, and nucleocapsid protein immunoglobulin (Ig)G and IgA antibodies with a laboratory-developed enzyme-linked immunosorbent assay and tested how antigen, isotype and threshold choices affected the seroprevalence outcomes. The following threshold methods were evaluated: (i) mean 3 standard deviations of the negative controls; (ii) 100 specificity for each antigen-isotype combination; and (iii) the maximal Youden index. RESULTS: We found vastly different seroprevalence estimates depending on selected antigens and isotypes and the applied threshold method, ranging from 0.0 to 85.4. Subsequently, we maximized specificity and reported a seroprevalence, based on more than one antigen, ranging from 9.3 to 25.9. CONCLUSIONS: This study revealed the importance of evaluating serosurvey tools for antigen-, isotype-, and threshold-specific sensitivity and specificity, to interpret qualitative serosurvey outcomes reliably and consistently across studies.

Alterations and diagnostic performance of capillary ketonemia in pediatric acute appendicitis: a pilot study.

INTRODUCTION: The diagnostic performance of capillary ketonemia (CK) has been previously evaluated in context of pediatric acute gastroenteritis. To our knowledge, there is no literature on its performance in the setting of pediatric acute appendicitis (PAA). MATERIALS AND METHODS: In this study, 151 patients were prospectively included and divided into two groups: (1) patients with non-surgical abdominal pain in whom the diagnosis of PAA was excluded (n = 53) and (2) patients with a confirmed diagnosis of PAA (n = 98). In 80 patients (Group 1, n = 23 and group 2, n = 57) a CK was measured at the time of diagnosis. The PAA group was further classified into complicated (n = 18) and uncomplicated PAA (n = 39). Quantitative variables were compared between groups using the Mann-Whitney U test. Diagnostic performance of CK was evaluated with ROC curves. RESULTS: CK values were 0.3 [0.1-0.9] mmol/L in group 1 and 0.7 [0.4-1.4] mmol/L in group 2 (p = 0.01). Regarding the type of PAA, CK values were 0.6 [0.4-0.9] mmol/L in uncomplicated PAA and 1.2 [0.8-1.4] mmol/L in complicated PAA (p = 0.02). The AUC for the discrimination between groups 1 and 2 was 0.68 (95% IC 0.53-0.82) (p = 0.24) and the AUC for the discrimination between uncomplicated PAA and complicated PAA was 0.69 (95% IC 0.54-0.85) (p = 0.04). The best cut-off point (group 1 vs group 2) resulted in 0.4 mmol/L, with a sensitivity of 80.7% and a specificity of 52.2%. The best cut-off point (non-complicated vs complicated PAA) resulted in 1.1 mmol/L, with a sensitivity of 61.1% and a specificity of 76.9%. CONCLUSIONS: This study found significantly higher levels of CK in patients with PAA than in those with NSAP. Similarly, significantly higher levels were observed in patients with complicated than in those with uncomplicated PAA. Nevertheless, the diagnostic performance of CK was only moderate in the two settings analyzed. The potential usefulness of CK determination as a tool to guide the preoperative rehydration regimen of patients with PAA to prevent postoperative hyporexia and vomiting is a promising line of research and should be evaluated in future studies.

Robust RNA-seq data analysis using an integrated method of ROC curve and Kolmogorov-Smirnov test.

It is a common approach to dichotomize a continuous biomarker in clinical setting for the convenience of application. Analytically, results from using a dichotomized biomarker are often more reliable and resistant to outliers, bi-modal and other unknown distributions. There are two commonly used methods for selecting the best cut-off value for dichotomization of a continuous biomarker, using either maximally selected chi-square statistic or a ROC curve, specifically the Youden Index. In this paper, we explained that in many situations, it is inappropriate to use the former. By using the Maximum Absolute Youden Index (MAYI), we demonstrated that the integration of a MAYI and the Kolmogorov-Smirnov test is not only a robust non-parametric method, but also provides more meaningful p value for selecting the cut-off value than using a Mann-Whitney test. In addition, our method can be applied directly in clinical settings.

Multi-center verification of the influence of data ratio of training sets on test results of an AI system for detecting early gastric cancer based on the YOLO-v4 algorithm.

Objective: Convolutional Neural Network(CNN) is increasingly being applied in the diagnosis of gastric cancer. However, the impact of proportion of internal data in the training set on test results has not been sufficiently studied. Here, we constructed an artificial intelligence (AI) system called EGC-YOLOV4 using the YOLO-v4 algorithm to explore the optimal ratio of training set with the power to diagnose early gastric cancer. Design: A total of 22,0918 gastroscopic images from Yixing People's Hospital were collected. 7 training set models were established to identify 4 test sets. Respective sensitivity, specificity, Youden index, accuracy, and corresponding thresholds were tested, and ROC curves were plotted. Results: 1. The EGC-YOLOV4 system completes all tests at an average reading speed of about 15 ms/sheet; 2. The AUC values in training set 1 model were 0.8325, 0.8307, 0.8706, and 0.8279, in training set 2 model were 0.8674, 0.8635, 0.9056, and 0.9249, in training set 3 model were 0.8544, 0.8881, 0.9072, and 0.9237, in training set 4 model were 0.8271, 0.9020, 0.9102, and 0.9316, in training set 5 model were 0.8249, 0.8484, 0.8796, and 0.8931, in training set 6 model were 0.8235, 0.8539, 0.9002, and 0.9051, in training set 7 model were 0.7581, 0.8082, 0.8803, and 0.8763. Conclusion: EGC-YOLOV4 can quickly and accurately identify the early gastric cancer lesions in gastroscopic images, and has good generalization.The proportion of positive and negative samples in the training set will affect the overall diagnostic performance of AI.In this study, the optimal ratio of positive samples to negative samples in the training set is 1:1~ 1:2.