Area Under ROC Curve Research Articles

Credit Risk Detection in Peer-to-Peer Lending Using CatBoost

P2P lending (Peer-to-peer lending) is widely used by private borrowers, small businesses, and MSMEs because P2P lending allows individuals and businesses to be able to lend money directly from lenders without the stringent requirements and criteria of traditional banks and financial institutions. However, P2P lending has a credit risk problem characterized by a high failure rate for borrowers to repay their loans. Therefore, a system was necessary to detect credit risk to minimize the risk of P2P lending. In this study, a system had been built using the CatBoost method; the dataset used was taken from the Bondora loan dataset. To measure the performance of the CatBoost algorithm, an evaluation matrix was performed using ROC (Receiver Operating Characteristics) curves and AUC (Area Under Curve) was performed. The experiment consists of three scenarios, of which the best result regards Scenario 2 with a data splitting of 90:10. It was caused by the result of AUC value 0.80329 compared to scenario 1 with a data split of 80:20 with the AUC value around 0.789583, and scenario 3 with a data split of 70:30 with the AUC value around 0.781066, respectively.

Leveraging Quantitative Imaging and Machine Learning to Differentiate Radionecrosis from Disease Recurrence in Patients with Brain Metastases

Radiation necrosis can be difficult to non-invasively discern from tumor progression after stereotactic radiosurgery (SRS). In this work, we investigate the utility of radiomics (computerized features) and machine learning to capture per-voxel lesion heterogeneity on routine MRI scans, to differentiate radionecrosis from tumor recurrence in patients with brain metastases treated with SRS. A retrospective analysis was conducted of patients with brain metastases treated with SRS. Eighty-three lesions (n = 56 intact; n = 27 surgical cavity) from 69 patients were identified with median age 68.8 years (range 40.2 - 91.0), of whom 53.6% were male and 33.3% received prior whole-brain radiotherapy (WBRT). Lesion histology included lung (60.2%), renal cell (15.7%), melanoma (10.8%), breast (9.6%), and other (3.6%). Pathologic confirmation was available in 73.5% of lesions. Both intact and resection cavity lesions were included and individually segmented. Image preprocessing and radiomic feature extraction were done using ANTsPy and open-source software. A total of 210 features were extracted from post-contrast T1-weighted (T1w) and T2/FLAIR MRIs. Highly correlated features were removed. Univariate logistic regression was conducted on the remaining T1w and T2/FLAIR features as well as on clinical variables. Multivariate analysis was implemented with various classifiers (Random Forest, Ridge, Lasso, Support Vector Machine [SVM]) on the top-performing features found on univariate logistic regression. Models were assessed using cross-validation to select the best model by area under ROC curve (AUC). Specificity and sensitivity were calculated. On univariate analysis, the top 10 radiomics features consisted of 6 T1w features and 4 T2/FLAIR features (4 GLCM, 3 first order, 1 GLSZM, 1 GLRLM, and 1 shape feature). Age, gender, disease site, prior WBRT, prior fractionated SRS, planning tumor volume, brain-GTV V12 Gy, and immunotherapy before or after SRS were not predictive (AUC less than 62.0%) on univariate analysis compared to radiomic features. Multivariate analysis of top performing radiomic features on both intact and surgical cavities yielded an AUC of 72.0% (standard deviation [SD] ±8.8%). Multivariate analysis of top features on intact lesions alone improved the AUC to 80.5% (SD ±10.8%), with sensitivity of 77.8%, specificity of 72.4%, and positive likelihood ratio of 2.82 in differentiating radionecrosis from recurrence. Radiomics and machine learning tools may improve diagnostic ability of distinguishing radiation necrosis from tumor recurrence after SRS. Further work is needed to deploy this in a larger multi-institutional cohort and prospectively evaluate its efficacy as a decision-support tool to personalize care in patients with brain metastases.

Clinico-Genomics Model to Predict Survival in Non-Metastatic Breast Cancer: Comparing the Performance of Machine Learning Models

Breast cancer is a heterogenous spectrum of disease with variable outcome. This heterogeneity of outcome could be driven by underlying heterogeneity in the genotypes. There remains an unmet need of defining prognosis based on underlying genetic heterogeneity. The primary objective of this study was to build a combined clinic-genomics model to predict overall survival (OS) in patients with non-metastatic breast cancer. The Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) database is a combined Canada-UK Project which contains information on 31 demographic, clinical and treatment characteristics in addition to m-RNA level z-score for 331 genes for 1904 breast cancer patients. We excluded patients with metaplastic breast cancer (n = 1) or unknown histologic subtype (n = 13), and those with missing information on breast surgery (n = 22) and cellularity (n = 54). Demographic characteristics included age of the patient and menopausal status while clinical characteristics included tumor stage, number of positive lymph nodes, pathologic grade, cellularity, receptor status for estrogen and progesterone receptors, and HER2neu gene amplification status. Treatment included type of breast cancer surgery, chemotherapy, hormone therapy, and radiation therapy. The data was split randomly at an 80:20 ratio into training and testing dataset. Subsequently, two cross-validated supervised machine learning algorithms were applied to build the prognostic model - elastic net logistic regression and random forest classifier. We compared the performance of the models in the testing dataset using area under receiver operating curve (AUC). All analysis were done using sparklyr - an R interface to Apache SparkTM. Overall, 1814 patients were included in the analysis. The AUC for elastic net logistic regression model and random forest classifier was 0.74, and 0.73, respectively. The sensitivity and specificity of the elastic net logistic regression model was 0.84 and 0.45, respectively while those for the random forest classifier was 0.83 and 0.51, respectively. STAT5A, CASP8, HSD17b11, and CDKN2c were common to the list of important genes predicting OS in both models while age, lymph node count, Nottingham Prognostic Index (NPI), and type of breast surgery were common to the list of clinical factors predicting OS in both models. This study shows that a multivariable prognostic model with combination of clinical and demographic parameters with Z-score expression data can predict for overall survival.

AI Serial Image Prediction of Progression-Free Survival (PFS) for Locally Advanced Non-Small Cell Lung Cancer (LA-NSCLC) Patients Treated with Chemoradiation (CRT) and Durvalumab Consolidation

Patient outcomes with definitive CRT for LA-NSCLC remain poor, with no imaging biomarkers to predict benefit. Hence, we developed a serial image AI model using paired planning CT (pCT) and first week cone-beam CT (CBCT) to predict PFS and measured AI model fairness defined as the bias in the classification with respect to gender as a protected attribute. Sixty-four consecutive patients with LA-NSCLC treated with concurrent CRT to 60 Gy in 30 fractions and durvalumab consolidation were analyzed. Three prediction models were created. A previously developed AI image foundation model [1] was pre-trained with unlabeled 6,402 3D CT scans sourced from institutional and the Cancer Imaging Archive and modified to predict PFS as a binarized outcome (high PFS > 6 months and low PFS < 6 months) using pCT scans. Serial image AI model was created by adding the first week CBCT scan. The third model measured tumor growth rate (TGR) as relative change in tumor and nodal volume from pCT to CBCT derived using a different published AI model [2]. Association with PFS using univariable and multivariable Cox regression after adjusting for age, gender, planning tumor volume, and smoking status were measured using TGR and the two AI model predictions using a cutoff of > 50% probability for low PFS. AI model fairness metrics area under receiver operating curve (AUROC), precision, sensitivity, and specificity were computed. TGR was not associated with PFS on univariate (Hazard ratio [HR] of 1.515, 95% confidence interval [CI] of 0.32 to 7.26, p = 0.60) or multivariate analysis (HR: 1.58, 95% CI: 0.32 to 7.80, p = 0.58) and resulted in a Harrell's C-index of 54.7%. The serial image AI model prediction was associated with PFS in both univariable (HR: 2.12, 95% CI: 1.02 to 4.40, p = 0.045) and multivariable analysis (HR 2.39, 95% CI of 1.09 to 5.25, p = 0.029), and a C-index of 62.5%. The pCT AI model was associated with PFS in univariate (HR 2.06, 95% CI of 1.06 to 4.01, p = 0.034) but not in multivariable analysis (HR 1.89, 95% CI of 0.93 to 3.87, p = 0.08), and a C-index of 59.9%. The serial image AI model reduced the parity in classification compared to pCT AI model indicating higher fairness (Table I). The multi-image AI model predicted PFS with slightly higher accuracy and resulted in higher fairness than the pCT AI model. These results underscore the potential for incorporating multi-imaging biomarkers to predict treatment response.

An Adaptive Multi-Feature Fusion Network for Predicting Overall Survival of Patients with Head and Neck Cancer

Accurate prognostic prediction could allow personalized treatment to achieve optimal clinical outcome. We aimed to develop a highly predictive overall survival model, considering the complementary relationships between clinical information, traditional radiomics and deep image information, to further improve the overall prediction accuracy by constructing a richer feature set and adaptive weighting. A total of 427 patients with Oropharyngeal Cancer (OPC) patients from the TCIA database were included. 341 cases were used for training, 86 cases were used as an independent cohort. Patient characteristics, including TMN, age, gender, HPV status, smoking or drinking status, etc. were considered as potential predictors. Traditional radiomics features of gross tumor volume (GTV) was extracted from planning CT using open-source software. In addition, a two-dimensional convolutional network (2D_CNN) was designed to extract deep image features. An adaptive multi-feature fusion network was developed to predict overall survival of patients based on three types of features. The fusion network integrates an attention mechanism to the channel dimension to obtain proper weighting of each channel in the feature graph through the fully connected network by focusing on effective feature channels and automatic learning according to the loss, thus improving the utilization rate of effective features. The model performance was evaluated using the area-under-ROC-curve (AUC), accuracy, precision, recall, f1-score. The AUCs of predictive models based on clinical features, traditional radiomics features and deep image features were 0.7, 0.61 and 0.72, respectively. Combining patient characteristics, radiomic features and deep imaging features, the AUCs of the prediction models was significantly improved to 0.85 and 0.86 (with attention mechanisms) for the independent test cohort (Table 1). The proposed adaptive multi-channel network assigned effective weights to the potential predictors, selectively enhanced useful features while suppressed irrelevant features, enabling more accurate feature map weights. We demonstrated the improved predictive value, with a multi-channel fusion network integrated with an attention mechanism, for overall survival of OPC patients.

Fatty liver is associated with significant liver inflammation and increases the burden of advanced fibrosis in chronic HBV infection

BackgroundConcurrent non-alcoholic fatty liver disease (NAFLD) is common in patients with chronic HBV infection. But the impact of fatty liver on the histologic progression of HBV infection remains controversial.MethodsConsecutive HBV-infected patients who underwent liver biopsy between 2016 and 2021 were included. Alcohol consumption and other types of viral hepatitis were excluded. All biopsies were scored for grading and staging by Scheuer’s score, and the steatosis was scored as an estimate of the percentage of liver parenchyma replaced by fat. Logistic regression analyses were applied to assess the associated factors for significant liver inflammation (G ≥ 2), significant fibrosis (S ≥ 2) and advanced fibrosis (S ≥ 3).ResultsAmong the 871 HBV-infected patients, hepatic steatosis was prevalent in 255 patients (29.28%). Significant liver inflammation was present in 461 patients (52.93%). Significant fibrosis was observed in 527 patients (60.51%), while advanced liver fibrosis was observed in 171 patients (19.63%). Patients with concomitant NAFLD were more likely to have significant liver inflammation and advanced fibrosis. Fatty liver was an independent risk factor for significant liver inflammation (OR: 2.117, 95% CI: 1.500-2.988), but it could not predict the development of fibrosis. Especially, in HBV-infected patients with persistent normal ALT (immune tolerant and inactive carrier phase), the presence of significant liver inflammation was higher in NAFLD than those without NAFLD. The prevalence of advanced liver fibrosis was higher in NAFLD than non-NAFLD only in the immune tolerant phase, while NAFLD did not increase fibrosis burden in other stages of HBV infection. We developed a predictive model for significant liver inflammation with the area under receiver operating characteristic curve (AUROC) of 0.825, and a model for significant fibrosis with the AUROC of 0.760.ConclusionsNAFLD is independently associated with significant liver inflammation, and increases the burden of advanced liver fibrosis in HBV-infected patients. The influence of NAFLD on the degree of liver inflammation and fibrosis is different in distinct clinical phases of chronic HBV infection.

Prior physical illness predicts death better than acute physiological derangement on intensive care unit admission in COVID-19: A Swedish registry study.

COVID-19 is associated with prolonged intensive care unit (ICU) stay and considerable mortality. The onset of persistent critical illness, defined as when prior illness predicts death better than acute physiological derangement, has not been studied in COVID-19. This national cohort study based on the Swedish Intensive Care Registry (SIR) included all patients admitted to a Swedish ICU due to COVID-19 from 6 March 2020 to 9 November 2021. Simplified Acute Physiology Score-3 (SAPS3) Box 1 was used as a measure of prior illness and Box 3 as a measure of acute derangement to evaluate the onset and importance of persistent critical illness in COVID-19. To compare predictive capacity, the area under receiver operating characteristic (AUC) of SAPS3 and its constituent Box 1 and 3 was calculated for 30-day mortality. In 7 969 patients, of which 1 878 (23.6%) died within 30 days of ICU admission, the complete SAPS3 score had acceptable discrimination: AUC 0.75 (95% CI 0.74 to 0.76) but showed under prediction in low-risk patients and over prediction in high-risk patients. SAPS3 Box 1 showed markedly better discrimination than Box 3 (AUC 0.74 vs 0.65, P<0,0001). Using custom logistic models, the difference in predictive performance of prior and acute illness was validated, AUC 0.76 vs AUC 0.69, p<0.0001. Prior physical illness predicts death in COVID-19 better than acute physiological derangement during ICU stay, and the whole SAPS3 score is not significantly better than just prior illness. The results suggests that COVID-19 may exhibit similarities to persistent critical illness immediately from ICU admission, potentially because of long median ICU length-of-stay. Alternatively, the variables in the acute physiological derangement model may not adequately capture the severity of illness in COVID-19.

Prediction of Maxillary Bone Invasion in Hard Palate/Upper Alveolus Cancer: A Multi-Center Retrospective Study.

maxillary bone invasion (MBI) is not uncommon in hard palate or upper alveolus (HP/UA) cancer; however, there have been relatively few reports about the MBI of HP/UA cancer. this was a multi-center retrospective study, enrolling 144 cases of HP/UA cancer. MBI was defined by surgical pathology or radiology follow-up. The multiple prediction models for MBI were developed in total cases and in cases having primary bone resection, using clinical and radiological variables. computerized tomography (CT) alone predicted MBI, with an area under receiver operating curve (AUC) of 0.779 (95% confidence interval (CI) = 0.712-0.847). The AUC was increased in a model that combined tumor dimensions and clinical factors (male sex and nodal metastasis) (0.854 (95%CI = 0.790-0.918)). In patients who underwent 18fluorodeoxyglucose positron emission tomography/CT (PET/CT), the discrimination performance of a model including the maximal standardized uptake value (SUVmax) had an AUC of 0.911 (95%CI = 0.847-0.975). The scoring system using CT finding, tumor dimension, and clinical factors, with/without PET/CT SUVmax clearly distinguished low-, intermediate-, and high-risk groups for MBI. using information from CT, tumor dimension, clinical factors, and the SUVmax value, the MBI of HP/UA cancer can be predicted with a relatively high discrimination performance.

Chest MRI and CT Predictors of 10-Year All-Cause Mortality in COPD

Pulmonary imaging measurements using magnetic resonance imaging (MRI) and computed tomography (CT) have the potential to deepen our understanding of chronic obstructive pulmonary disease (COPD) by measuring airway and parenchymal pathologic information that cannot be provided by spirometry. Currently, MRI and CT measurements are not included in mortality risk predictions, diagnosis, or COPD staging. We evaluated baseline pulmonary function, MRI and CT measurements alongside imaging texture-features to predict 10-year all-cause mortality in ex-smokers with (n = 93; 31 females; 70 ± 9years) and without (n = 69; 29 females, 69 ± 9years) COPD. CT airway and vessel measurements, helium-3 (3He) MRI ventilation defect percent (VDP) and apparent diffusion coefficients (ADC) were quantified. MRI and CT texture-features were extracted using PyRadiomics (version2.2.0). Associations between 10-year all-cause mortality and all clinical and imaging measurements were evaluated using multivariable regression model odds-ratios. Machine-learning predictive models for 10-year all-cause mortality were evaluated using area-under-receiver-operator-characteristic-curve (AUC), sensitivity and specificity analyses. DLCO (%pred) (HR = 0.955, 95%CI: 0.934-0.976, p < 0.001), MRI ADC (HR = 1.843, 95%CI: 1.260-2.871, p < 0.001), and CT informational-measure-of-correlation (HR = 3.546, 95% CI: 1.660-7.573, p = 0.001) were the strongest predictors of 10-year mortality. A machine-learning model trained on clinical, imaging, and imaging textures was the best predictive model (AUC = 0.82, sensitivity = 83%, specificity = 84%) and outperformed the solely clinical model (AUC = 0.76, sensitivity = 77%, specificity = 79%). In ex-smokers, regardless of COPD status, addition of CT and MR imaging texture measurements to clinical models provided unique prognostic information of mortality risk that can allow for better clinical management.Clinical Trial Registration: www.clinicaltrials.gov NCT02279329

Application of machine learning model in predicting the likelihood of blood transfusion after hip fracture surgery.

Anemia is one of the common adverse reactions after hip fracture surgery. The traditional method to solve anemia is allogeneic transfusion. However, the transfusion may lead to some complications such as septicemia and fever. So far, few studies have reported roles of machine learning in predicting whether blood transfusion is needed or not after hip fracture surgery. Therefore, the purpose of this study is to develop machine learning models to predict the likelihood of postoperative blood transfusion in patients undergoing hip fracture surgery. This study enrolled 1355 patients who underwent hip fracture surgery at the Affiliated Hospital of Qingdao University from January 2016 to December 2021. Among all patients, 210 cases received postoperative blood transfusion. All patients were randomly divided into a training group and a testing group at a ratio of 7:3. In the training group, univariate and multivariate logistic regression analyses were used to determine independent risk factors for the postoperative transfusion. Then, based on these independent risk factors, tenfold cross-validation method was utilized to develop five machine learning models, including logistic, multilayer perceptron (MLP), extreme gradient boosting (XGBoost), random forest (RF), and support vector machine (SVM). The receiver operating characteristic (ROC) curve, area under ROC curve (AUC), and Matthews correlation coefficient (MCC) were generated to evaluate the performance of the models. Calibration plot and decision curve analysis (DCA) were used to test the performance, stability, and clinical applicability of the models. The models were validated using the testing group; and the ROC curve, MCC, calibration plot, and DCA curves were also generated to validate the performance, stability, and clinical applicability of the models. To further verify the robustness of the model, we randomly grabbed 70% of the samples in the testing set, performed 1000 iterations, and calculated the AUC and confidence interval of the five models. Finally, we used SHapley Additive exPlanations (SHAP) to explain these models. Multivariate logistic regression analysis showed that there were 8 independent risk factors, including age, blood transfusion history, albumin (ALB), globulin (GLO), total bilirubin (TBIL), indirect bilirubin (IBIL), hemoglobin (HB), and blood loss > 200ml. We finally selected five independent risk factors including HB, GLO, age, IBIL, and blood loss > 200ml. Based on these five independent risk factors, we generated six characteristic variables, namely HB, HB × HB, HB × blood loss, GLO × HB, age, age × IBIL, and established five machine learning models using a tenfold cross-validation method. In the training group, the AUC values of logistic, RF, MLP, SVM, and XGB were 0.9320, 0.8911, 0.9327, 0.9225, and 0.8825, respectively, and the average AUC was 0.9122 ± 0.0212. The MCC values were 0.65, 0.77, 0.65, 0.66, and 0.68, respectively, and the calibration plot and DCA performed well. In the testing group the AUC values of logistic, RF, MLP, SVM, and XGB were 0.8483, 0.7978, 0.8576, 0.8598, and 0.8216, respectively. The average AUC was 0.8370 ± 0.0238, and the MCC values were 0.41, 0.35, 0.40, 0.41, and 0.41, respectively. The calibration plot and DCA in the testing group also showed good performance. The AUC values and confidence intervals of the 1000-iteration model were: logistic (AUC, min confidence interval [CI]-max confidence interval [CI] 0.848, 0.804-0.903), RF (AUC, minCI-maxCI 0.797, 0.734-0.857), MLP (AUC, minCI-maxCI 0.858, 0.812-0.902), SVM (AUC, minCI-maxCI 0.859, 0.819-0.910), and XGB (AUC, minCI-maxCI 0.821, 0.764-0.894). The model performed well. Finally, according to SHAP, among all five models, HB played the most important role in model prediction and interpretation. The five models we developed all performed well in predicting the likelihood of blood transfusion after hip fracture surgery. Therefore, we believed that the prediction model based on machine learning had great application prospects in clinical practice, which could help clinicians better predict the risk of blood transfusion after hip fracture surgery.

An effective disease diagnostic model related to pyroptosis in ischemic cardiomyopathy.

Pyroptosis is involved in ischemic cardiomyopathy (ICM). The study aimed to investigate the pyroptosis-related genes and clarify their diagnostic value in ICM. The bioinformatics method identified the differential pyroptosis genes between the normal control and ICM samples from online datasets. Then, protein-protein interaction (PPI) and function analysis were carried out to explore the function of these genes. Following, subtype analysis was performed using ConsensusClusterPlus, functions, immune score, stromal score, immune cell proportion and human leukocyte antigen (HLA) genes between subtypes were investigated. Moreover, optimal pyroptosis genes were selected using the least absolute shrinkage and selection operator (LASSO) analysis to construct a diagnostic model and evaluate its effectiveness using receiver operator characteristic (ROC) analysis. Twenty-one differential expressed pyroptosis genes were identified, and these genes were related to immune and pyroptosis. Subtype analysis identified two obvious subtypes: sub-1 and sub-2. And LASSO identified 13 optimal genes used to construct the diagnostic model. The diagnostic model in ICM diagnosis with the area under ROC (AUC) was 0.965. Our results suggested that pyroptosis was tightly associated with ICM.

An Ensemble Models for the Prediction of Sickle Cell Disease from Erythrocytes Smears

INTRODUCTION: The human blood as a collection of tissues containing Red Blood Cells (RBCs), circular in shape and acting as an oxygen carrier, are frequently deformed by multiple blood diseases inherited from parents. These hereditary diseases of blood involve abnormal haemoglobin (Hb) or anemia which are major public health issues. Sickle Cell Disease (SCD) is one of the common non-communicable disease and genetic disorder due to changes in hematological conditions of the RBCs which often causes the inheritance of mutant Hb genes by the patient..&#x0D; OBJECTIVES: The process of manual valuation, predictions and diagnosis of SCD necessitate for a passionate time spending and if not done properly can lead to wrong predictions and diagnosis. Machine Learning (ML), a branch of AI which emphases on building systems that improve performance based on the data they consume is appropriate. Despite previous research efforts in predicting with single ML algorithm, the existing systems still suffer from high false and wrong predictions.&#x0D; METHODS: Thus, this paper aimed at performing comparative analysis of individual ML algorithms and their ensemble models for effective predictions of SCD (elongated shapes) in erythrocytes blood cells. Three ML algorithms were selected, and ensemble models were developed to perform the predictions and metrics were used to evaluate the performance of the model using accuracy, sensitivity, Receiver Operating Characteristics-Area under Curve (ROC-AUC) and F1 score metrics. The results were compared with existing literature for model(s) with the best prediction metrics performance..&#x0D; RESULTS: The analysis was carried out using Python programming language. Individual ML algorithms reveals that their accuracies show MLR=87%, XGBoost=90%, and RF=93%, while hybridized RF-MLR=92% and RF-XGBoost=99%. The accuracy of RF-XGBoost of 99% outperformed other individual ML algorithms and Hybrid models. &#x0D; CONCLUSION: Thus, the study concluded that involving hybridized ML algorithms in medical datasets increased predictions performance as it removed the challenges of high variance, low accuracy and feature noise and biases of medical datasets. The paper concluded that ensemble classifiers should be considered to improve sickle cell disease predictions.

Utility of Contrast-Enhanced Computed Tomography for Differentiating Pancreatic Ductal Adenocarcinoma from Inflammatory Mass in Chronic Calcific Pancreatitis

Abstract Objective The aim of this study was to identify the most useful contrast-enhanced computed tomography (CECT) features for differentiating pancreatic ductal adenocarcinoma (PDAC) from mass-forming chronic pancreatitis (MFCP) in chronic calcific pancreatitis (CCP). Methods In total, 101 patients with CCP and focal pancreatic mass formed the study group. Sixteen qualitative and four quantitative parameters were analyzed. Qualitative parameters included size, site, margin, intralesional hypodensity, collateral duct sign, abrupt pancreatic duct (PD) cutoff, upstream PD dilatation, distal pancreatic atrophy, double duct sign, enhancement pattern, contrast attenuation, peripancreatic inflammation, vascular involvement, regional nodes, and metastasis. Quantitative parameters included duct-to-body ratio, common bile duct (CBD) diameter, main pancreatic duct (MPD) diameter, and carcinoembryonic antigen 19-9 (CA19-9). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated for prediction of PDAC. A receiver operating characteristic (ROC) curve analysis was performed and the area under ROC curve (AUROC) was calculated to determine diagnostic accuracy to assess the optimal cutoff. Results PDAC was confirmed in 48 patients and MFCP in 53 patients. A duct-to-body ratio greater than 0.48 had 95.5% sensitivity, 83.3% specificity, 80.8% PPV, 96.2% NPV, and 88.5% accuracy for predicting PDAC. A CBD diameter cutoff ≥9.5 mm had an accuracy of 75% (p &lt; 0.019) and an MPD cutoff ≥6.25 mm had an accuracy of 67.8% (p = 0.008) for predicting PDAC. On binary logistic regression, the duct-to-body ratio was found to be the significant independent factor associated with malignancy. Conclusion A duct-to-body ratio greater than 0.48, intralesional hypodensity, and abrupt duct cutoff are the most helpful computed tomography (CT) features for distinguishing PDAC from MFCP in CCP. On binary logistic regression, the duct-to-body ratio was found to be a significant independent factor. Interspersed normal parenchyma was observed as a very specific sign of MFCP. Intraparenchymal hypodensity has high specificity, but further validation is needed.

Pleural fluid adenosine deaminase to serum C-reactive protein ratio for diagnosing tuberculous pleural effusion

BackgroundTuberculous pleural effusion (TPE) and malignant pleural effusion (MPE) may occasionally show similar cytological and biochemical picture including ADA. In such cases, differentiating TPE and MPE is challenging and needs histopathology of pleural tissue which may involve invasive procedures. The present study aims to evaluate the diagnostic accuracy of pleural fluid ADA to serum CRP (ADA/CRP) ratio to discriminate between tuberculous and malignant pleural effusion. In addition, we investigated whether the ratio ADA/CRP adds diagnostic value to ADA.MethodsThis cross-sectional study was conducted in the National Institute of Diseases of the Chest and Hospital (NIDCH), Mohakhali, Dhaka, from July 2021 to February 2022 on diagnosed patients of TPE and malignant pleural effusion MPE. A receiver operating characteristic curve (ROC) was constructed for identifying TPE. The added value of the ADA/CRP ratio to ADA was evaluated using the net reclassification improvement (NRI) and integrated discrimination improvement (IDI). A value of p < 0.05 was considered statistically significant for all tests.ResultsFifty-nine patients were enrolled in this study, of which 31 had TPE, and 28 had MPE. Pleural fluid ADA to serum CRP ratio and pleural fluid ADA level was significantly higher in patients with TPE, but there was no significant difference in serum CRP levels between patients with TPE and MPE. At cut off value of > 1.25, pleural fluid ADA to serum CRP ratio had a sensitivity of 93.8%, specificity of 85.2%, and positive and negative predictive values were 88.2% and 92% respectively, in the diagnosis of TPE and area under ROC curve (AUC) was 0.94. The NRI and IDI analyses revealed added diagnostic value of ADA/CRP to ADA.ConclusionThis study shows that the ADA/CRP ratio improves the diagnostic usefulness of ADA for TPE.

A retrospective study of myelin oligodendrocyte glycoprotein antibody-associated disease from a clinical laboratory perspective.

To analyze the differences in laboratory data between patients with myelin oligodendrocyte glycoprotein (MOG) antibody-associated disease (MOGAD), multiple sclerosis (MS) and neuromyelitis optica spectrum disorder (NMOSD). The study included 26 MOGAD patients who visited Beijing Tiantan Hospital from 2018 to 2021. MS and NMOSD patients who visited the clinic during the same period were selected as controls. Relevant indicators were compared between the MOGAD group and the MS/NMOSD groups, and the diagnostic performance of meaningful markers was assessed. The MOGAD group showed a slight female preponderance of 57.7%, with an average onset age of 29.8 years. The absolute and relative counts of neutrophils were higher in the MOGAD group than in the MS group, while the proportion of lymphocytes was lower. The cerebrospinal fluid (CSF) IgG level, IgG index, 24-h IgG synthesis rate, and positive rate of oligoclonal bands (OCB) were lower in MOGAD patients than in the MS group. The area under ROC curve (AUC) was 0.939 when combining the relative lymphocyte count and IgG index. Compared to the NMOSD group, the MOGAD group had higher levels of serum complement C4 and lower levels of serum IgG. The AUC of serum C4 combined with FT4 was 0.783. Statistically significant markers were observed in the laboratory data of MOGAD patients compared to MS/NMOSD patients. The relative lymphocyte count combined with IgG index had excellent diagnostic efficacy for MOGAD and MS, while serum C4 combined with FT4 had better diagnostic efficacy for MOGAD and NMOSD.

Distributed One-Pass Online AUC Maximization

AUC (Area Under ROC curve) is a widely used metric for measuring classification performance. Recently, due to the efficiency and scalability of online learning in handling large-scale problems, it attracts lots of interests to maximize AUC in online setting. Traditional online learning algorithms require a computation node to process all the data sequentially. However, due to the explosion in volume and velocity of modern data generated from distributed sources such as sensor network and Internet, it is not efficient to send all the data to one node for processing and updating model sequentially. To address this challenge, we explore the distributed learning techniques for maximizing AUC in online setting with a linear model. Specifically, we propose two different algorithms for solving the distributed online AUC maximization problem: (i) the Centralized Distributed One-Pass Online AUC Maximization (C-DOPOAM) algorithm in which the server updates the global model with the gradients computed from workers; and (ii) the Decentralized Distributed One-Pass Online AUC Maximization (D-DOPOAM) algorithm in which each worker updates its local model and exchanges information with its neighbors to achieve consensus. Moreover, we establish the regret bounds of the two proposed algorithms. They achieve the similar convergence rate with the state-of-the-art non-distributed online AUC maximization algorithms. Finally, we conduct extensive experiments to validate our theoretical result on different computation platforms and network configurations. We also make a comparison between C-DOPOAM and D-DOPOAM in poor network condition. The results show D-DOPOAM is more robust than C-DOPOAM against low bandwidth or high latency.

Using Machine Learning to Predict Bleeding after Cardiac Surgery.

The primary objective was to predict bleeding after cardiac surgery with machine learning using the data from the ANZSCTS (Australia New Zealand Society of Cardiac and Thoracic Surgeons) Cardiac Surgery Database, cardiopulmonary bypass perfusion database, ICU (intensive care unit) database and laboratory results. We obtained surgical, perfusion, ICU and laboratory data from a single Australian tertiary cardiac surgical hospital from February 2015 to March 2022 and included 2000 patients undergoing cardiac surgery. We trained our models to predict either the Papworth definition or Dyke et al.'s universal definition of perioperative bleeding. Our primary outcome was the performance of our machine learning algorithms using sensitivity, specificity, positive and negative predictive values, accuracy, AUROC (area under receiver operating characteristics curve) and AUPRC (area under precision recall curve). Of the 2000 patients undergoing cardiac surgery, 13.3% (226/2000) had bleeding using the Papworth definition and 17.2% (343/2000) had moderate to massive bleeding using Dyke et al.'s definition. The best performing model based on AUPRC was the Ensemble Voting Classifier model for both Papworth (AUPRC 0.310, AUROC 0.738) and Dyke definitions of bleeding (AUPRC 0.452, AUROC 0.797). Machine learning can incorporate routinely collected data from various datasets to predict bleeding after cardiac surgery.

Vertebral CT attenuation outperforms standard clinical fracture risk prediction tools in detecting osteoporotic disease in lung cancer screening participants.

Compare accuracy of vertebral Hounsfield Unit (VHU) attenuation and FRAX and Garvan Fracture Risk Calculators in identifying low bone mineral density (BMD) and prevalent vertebral compression fractures (VF) in lung cancer screening (LCS) participants. Baseline CT scans from a single site of the International Lung Screen Trial were analysed. BMD was measured using VHU (of the most caudally imaged vertebra) and quantitative CT (QCT) (low BMD defined as <110 HU and <120 mg/cm3, respectively). Prevalent VF were classified semi-quantitatively. 10-year FRAX and Garvan fracture risks were calculated using dual energy X-ray absorptiometry (DXA) femoral neck T-score where available. Discrimination was assessed by area under receiver-operating characteristic curves (AUC). 535 LCS participants were included; 41% had low VHU-BMD, 56% had low QCT-BMD and 10% had ≥1 VF with ≥25% vertebral height loss. VHU demonstrated 94% specificity and 70% sensitivity in identifying low QCT-BMD. VHU was superior to fracture risk tools in discriminating low QCT-BMD (AUC: VHU 0.94 vs FRAX 0.67, Garvan 0.64 [p < 0.05]). In 64 participants with recent DXA scans, VHU was superior to FRAXT-score and GarvanT-score in discriminating low QCT-BMD (AUC: VHU 0.99, FRAXT-score 0.71, GarvanT-score 0.71 [p < 0.05]). VHU was non-inferior to FRAXT-score and GarvanT-score in discriminating VF (AUC: VHU 0.65, FRAXT-score 0.53, GarvanT-score 0.61). VHU outperforms clinical risk calculators in detecting low BMD and discriminates prevalent VF equally well as risk calculators with T-scores, yet is significantly simpler to perform. VHU measurement could aid osteoporosis assessment in high-risk smokers undergoing LCS.

Artificial intelligence system-based histogram analysis of computed tomography features to predict tumor invasiveness of ground-glass nodules.

The use of an artificial intelligence (AI)-based diagnostic system can significantly aid in analyzing the histogram of pulmonary nodules. The aim of our study was to evaluate the value of computed tomography (CT) histogram indicators analyzed by AI in predicting the tumor invasiveness of ground-glass nodules (GGNs) and to determine the added value of contrast-enhanced CT (CECT) compared with nonenhanced CT (NECT) in this prediction. This study enrolled patients with persistent GGNs who underwent preoperative NECT and CECT scanning. AI-based histogram analysis was performed for pathologically confirmed GGNs, which was followed by screening invasiveness-related factors via univariable analysis. Multivariable logistic models were developed based on candidate CT histogram indicators measured on either NECT or CECT. Receiver operating characteristic (ROC) curve and precision-recall (PR) curve were used to evaluate the models' performance. A total of 116 patients comprising 121 GGNs were included and divided into the precancerous lesion and adenocarcinoma groups based on invasiveness. In the AI-based histogram analysis, the mean CT value [NECT: odds ratio (OR) =1.009; 95% confidence interval (CI): 1.004-1.013; P<0.001] and solid component volume (NECT: OR =1.005; 95% CI: 1.000-1.010; P=0.032) were associated with the adenocarcinoma and used for multivariable logistic modeling. The area under ROC curve (AUC) and PR curve (AUPR) were not significantly different between the NECT model (AUC =0.765, 95% CI: 0.679-0.837; AUPR =0.907, 95% CI: 0.825-0.953) and the optimal CECT model (delayed phase: AUC =0.772, 95% CI: 0.687-0.843; AUPR =0.895, 95% CI: 0.812-0.944). No significantly different metrics were observed between the NECT and CECT models (precision: 0.707 vs. 0.742; P=0.616). The AI diagnostic system can help in the diagnosis of GGNs. The system displayed decent performance in GGN detection and alert to malignancy. Mean CT value and solid component volume were independent predictors of tumor invasiveness. CECT provided no additional improvement in diagnostic performance as compared with NECT.

Abstract P347: Utility Of Standing Blood Pressure On Blood Pressure Control In Treated Hypertension

Many hypertension (HTN) guidelines have recommended routine measurement of standing BP in HTN adults at risk of orthostatic hypotension, especially in older adults. However, the incremental value of adding standing BP to seated BP in predicting HTN control has not been studied. Accordingly, we assessed BP in both seated and standing positions in 319 HTN adults treated with antihypertensive medications. Adequate BP control was defined as 24-Hr ABPM ≤125/75 or daytime ABPM ≤130/80 mmHg. Area under receiver operator characteristic curve (AUROC) was used to determine accuracy of seated and standing BP in assessing BP control. Bayes factor (BF) was used to assess the significant difference between AUROCs. Sensitivity and specificity of standing BP were derived using cutoffs derived from Youden’s Index. Overall mean age was 64±13 years, with 55% women (176 of 319), and 27% Black (86 of 319). Sensitivity and specificity of seated SBP were 75% and 60%, respectively, when HTN control was defined by 24-Hr ABPM. Optimal cutoffs for standing BP were 138/86 and 142/86 mmHg (based on 24-Hr SBP and daytime SBP as gold standard, respectively). Sensitivity and specificity of standing SBP were 54% and 81%, respectively, when using the cutoff derived from 24-Hr ABPM. The AUROCs of standing SBP and DBP were not significantly different compared to seated when HTN control was defined by 24-Hr or daytime ABPM. Similarly, the addition of standing to seated BP was not significantly different when compared to seated BP based on both references (Figure 1: A-B). In conclusion, routine incorporation of standing office BP did not offer additional value in determining adequacy of BP control among treated HTN patients.