Receiver Operating Characteristic AUC Research Articles

Predicting an unfavorable outcome in patients with liver failure associated with obstructive jaundice syndrome: a prospective observational study

INTRODUCTION: Specialized scales developed to assess survival and mortality of patients after liver transplantation have not been validated when applied to liver failure in the setting of obstructive jaundice. OBJECTIVE: To determine the prognostic value of multiparametric scoring scales and to identify predictors of unfavorable outcome in liver failure due to obstructive jaundice. MATERIALS AND METHODS: 53 patients with confirmed liver failure due to obstructive jaundice of benign etiology who underwent biliary decompression were included in this prospective, observational, single-center study. Depending on the outcome two groups were distinguished: with favorable (n = 27, 51 %) and with unfavorable outcome (n = 26, 49 %). The following time points were defined for the dynamic evaluation of the course of the disease: day 1, day 3, day 7, and day 11 after decompression. Statistical processing of the data was performed using IBM SPSS Statistics 22 software for social science data analysis. RESULTS: The sensitivity and specificity of the studied scales in predicting liver failure due to obstructive jaundice were calculated on the basis of receiver operating characteristic (ROC) analysis. The AUROC for the SOFA was 0.862, for the MELD — 0.882, for the APACHE II — 0.864, for the Child-Turcotte-Pugh — 0.813. The sensitivity and specificity values for the SOFA were 78.9 and 77.8 %, for the MELD — 80.8 and 79.7 %, for the APACHE II — 57.7 and 81.9 %, and for the Child-Turcotte-Pugh — 75.1 and 70.9 %, respectively. Albumin was an independent predictor of unfavorable outcome (AUC ROC 0.909, p = 0.01), sensitivity 88.6 %, specificity 85.2 %. CONCLUSIONS: The scales presented in the study and the biomarker candidate “albumin” showed significant prognostic ability, but low values of sensitivity and specificity (less than 80 %) in some points of the study require the search for additional predictors of unfavorable outcome of liver failure due to obstructive jaundice.

Serum uric acid level is associated with glomerular ischemic lesions in patients with primary membranous nephropathy: an analytical, cross-sectional study

To investigate the relationship between serum uric acid level and glomerular ischemic lesions (GIL) in patients with primary membranous nephropathy (PMN) and identify relevant risk factors. A total of 201 patients with PMN but normal renal function confirmed by renal biopsy executed in the Liaocheng People’s Hospital, China, during January 2020-January 2023 were analyzed retrospectively. The enrolled patients were divided into a hyperuricemia group and a normal serum uric acid group (control group) according to their serum uric acid levels. Then, the participants were further divided into a non-GIL group or a GIL group based on the patient’s renal biopsy results. The two groups’ clinical and pathological data and meaningful indicators for differences were analyzed by binary logistic regression analysis. Additionally, the serum uric acid level prediction value on GIL was investigated using receiver operating characteristic (ROC) curves. Compared with the control group, the hyperuricemia group exhibited high serum uric acid, the prevalence of GIL, serum albumin, the prevalence of hypertension, and low-density lipoprotein cholesterol (LDL) levels (P < 0.05). Compared with the non-GIL group, the GIL group exhibited were older, had enhanced serum uric acid, serum albumin, and an increased prevalence of tubular atrophy/interstitial fibrosis (TA/IF), arteriolosclerosis, and low eGFR levels (P < 0.05). The binary logistic regression analysis revealed that the serum uric acid and the TA/IF are independent risk factors of GIL (P < 0.05). The AUC of ROC of GIL of PMN patients, predicted based on the serum uric acid concentration, was 0.736 (P < 0.05), wherein the threshold = 426.5 μmol/L and the Youden’s index = 0.41. Serum uric acid concentration and the TA/IF are independent risk factors of GIL in patients with PMN, and the former exhibits prediction value on GIL in patients with PMN.

Development and external validation of a novel score for predicting postoperative 30‑day mortality in tumor craniotomy patients: A cross‑sectional diagnostic study.

The identification of patients with craniotomy at high risk for postoperative 30-day mortality may contribute to achieving targeted delivery of interventions. The present study aimed to develop a personalized nomogram and scoring system for predicting the risk of postoperative 30-day mortality in such patients. In this retrospective cross-sectional study, 18,642 patients with craniotomy were stratified into a training cohort (n=7,800; year of surgery, 2012-2013) and an external validation cohort (n=10,842; year of surgery, 2014-2015). The least absolute shrinkage and selection operator (LASSO) model was used to select the most important variables among the candidate variables. Furthermore, a stepwise logistic regression model was established to screen out the risk factors based on the predictors chosen by the LASSO model. The model and a nomogram were constructed. The area under the receiver operating characteristic (ROC) curve (AUC) and calibration plot analysis were used to assess the model's discrimination ability and accuracy. The associated risk factors were categorized according to clinical cutoff points to create a scoring model for postoperative 30-day mortality. The total score was divided into four risk categories: Extremely high, high, intermediate and low risk. The postoperative 30-day mortality rates were 2.43 and 2.58% in the training and validation cohort, respectively. A simple nomogram and scoring system were developed for predicting the risk of postoperative 30-day mortality according to the white blood cell count; hematocrit and blood urea nitrogen levels; age range; functional health status; and incidence of disseminated cancer cells. The ROC AUC of the nomogram was 0.795 (95% CI: 0.764 to 0.826) in the training cohort and it was 0.738 (95% CI: 0.7091 to 0.7674) in the validation cohort. The calibration demonstrated a perfect fit between the predicted 30-day mortality risk and the observed 30-day mortality risk. Low, intermediate, high and extremely high risk statuses for 30-day mortality were associated with total scores of (-1.5 to -1), (-0.5 to 0.5), (1 to 2) and (2.5 to 9), respectively. A personalized nomogram and scoring system for predicting postoperative 30-day mortality in adult patients who underwent craniotomy were developed and validated, and individuals at high risk of 30-day mortality were able to be identified.

Validation of a Stenotrophomonas maltophilia bloodstream infection prediction score in the hematologic malignancy population

Stenotrophomonas maltophilia (SM) bloodstream infections (BSIs) contribute to significant mortality in hematologic malignancy (HM) and hematopoietic stem cell transplantation (HSCT) patients. A risk score to predict SM BSI could reduce time to appropriate antimicrobial therapy (TTAT) and improve patient outcomes. A single center cohort study of hospitalized adults with HM/HSCT was conducted. Patients had ≥ 1 blood culture with a Gram-negative (GN) organism. A StenoSCORE was calculated for each patient. The StenoSCORE2 was developed using risk factors for SM BSI identified via logistic regression. Receiver operating characteristic (ROC) curves were plotted. Sensitivity and specificity for the StenoSCORE and StenoSCORE2 were calculated. Thirty-six SM patients and 534 non-SM patients were assessed. A StenoSCORE ≥ 33 points was 80% sensitive, 68% specific, and accurately classified 69% of GN BSIs. StenoSCORE2 variables included acute leukemia, prolonged neutropenia, mucositis, ICU admission, recent meropenem and/or cefepime exposure. The StenoSCORE2 performed better than the StenoSCORE (ROC AUC 0.84 vs. 0.77). A StenoSCORE2 ≥ 4 points was 86% sensitive, 76% specific, and accurately classified 77% of GN BSIs. TTAT was significantly longer for patients with SM BSI compared with non-SM BSI (45.16 h vs. 0.57 h; p < 0.0001). In-hospital and 28-day mortality were significantly higher for patients with SM BSI compared to non-SM BSI (58.3% vs. 18.5% and 66.7% vs. 26.4%; p-value < 0.0001). The StenoSCORE and StenoSCORE2 performed well in predicting SM BSIs in patients with HM/HSCT and GN BSI. Clinical studies evaluating whether StenoSCORE and/or StenoSCORE2 implementation improves TTAT and clinical outcomes are warranted.

DeepBiomarker2: Prediction of Alcohol and Substance Use Disorder Risk in Post-Traumatic Stress Disorder Patients Using Electronic Medical Records and Multiple Social Determinants of Health.

Prediction of high-risk events amongst patients with mental disorders is critical for personalized interventions. We developed DeepBiomarker2 by leveraging deep learning and natural language processing to analyze lab tests, medication use, diagnosis, social determinants of health (SDoH) parameters, and psychotherapy for outcome prediction. To increase the model's interpretability, we further refined our contribution analysis to identify key features by scaling with a factor from a reference feature. We applied DeepBiomarker2 to analyze the EMR data of 38,807 patients from the University of Pittsburgh Medical Center diagnosed with post-traumatic stress disorder (PTSD) to determine their risk of developing alcohol and substance use disorder (ASUD). DeepBiomarker2 predicted whether a PTSD patient would have a diagnosis of ASUD within the following 3 months with an average c-statistic (receiver operating characteristic AUC) of 0.93 and average F1 score, precision, and recall of 0.880, 0.895, and 0.866 in the test sets, respectively. Our study found that the medications clindamycin, enalapril, penicillin, valacyclovir, Xarelto/rivaroxaban, moxifloxacin, and atropine and the SDoH parameters access to psychotherapy, living in zip codes with a high normalized vegetative index, Gini index, and low-income segregation may have potential to reduce the risk of ASUDs in PTSD. In conclusion, the integration of SDoH information, coupled with the refined feature contribution analysis, empowers DeepBiomarker2 to accurately predict ASUD risk. Moreover, the model can further identify potential indicators of increased risk along with medications with beneficial effects.

Type-2 diabetes mellitus remission prediction models after Roux-En-Y gastric bypass and sleeve gastrectomy based on disease severity scores

AimMetabolic and bariatric surgery (MBS) is considered one of the most effective interventions for weight loss and associated type-2 diabetes mellitus (T2DM) remission. Multiple scores including the Individualized Metabolic Surgery (IMS), DiaRem, advanced DiaRem, and Robert et al. scores, have been developed predict T2DM remission after MBS. We aim to validate each of these scores in our cohort of patients undergoing MBS with long-term follow-up and assess their efficacy based on procedure type and preoperative BMI. MethodsWe conducted a multicenter cohort study including patients with T2DM undergoing either Roux-en-Y gastric bypass (RYGB) or sleeve gastrectomy (SG). Areas under the receiver operating characteristics (ROC) curve (AUC) were calculated to assess the discriminatory ability of the four models to detect T2DM remission. ResultsA total of 503 patients (67 % females, mean age 53.5 [11] years, BMI 46.2 [8.8] kg/m2) with T2DM were included. The majority (78 %) underwent RYGB, while the rest (28 %) had SG. All four scores predicted T2DM remission in our cohort with an ROC AUC of 0.79 for IMS, 0.78 for both DiaRem and advanced-DiaRem, and 0.75 for Robert et al. score. Specific subgroups for each of these scores demonstrated higher T2DM remission rates after RYGB compared to SG. ConclusionWe demonstrate the ability of the IMS, DiaRem, advanced-DiaRem and Robert et al. scores to predict T2DM remission in patients undergoing MBS. T2DM remission rates was demonstrated to decrease with more severe IMS, DiaRem and advanced-DiaRem scores and lower Robert et al. scores.

Metabolic imaging of human embryos is predictive of ploidy status but is not associated with clinical pregnancy outcomes: a pilot trial.

Does fluorescence lifetime imaging microscopy (FLIM)-based metabolic imaging assessment of human blastocysts prior to frozen transfer correlate with pregnancy outcomes? FLIM failed to distinguish consistent patterns in mitochondrial metabolism between blastocysts leading to pregnancy compared to those that did not. FLIM measurements provide quantitative information on NAD(P)H and flavin adenine dinucleotide (FAD+) concentrations. The metabolism of embryos has long been linked to their viability, suggesting the potential utility of metabolic measurements to aid in selection. This was a pilot trial enrolling 121 IVF couples who consented to have their frozen blastocyst measured using non-invasive metabolic imaging. After being warmed, 105 couples' good-quality blastocysts underwent a 6-min scan in a controlled temperature and gas environment. FLIM-assessed blastocysts were then transferred without any intervention in management. Eight metabolic parameters were obtained from each blastocyst (4 for NAD(P)H and 4 for FAD): short and long fluorescence lifetime, fluorescence intensity, and fraction of the molecule engaged with enzyme. The redox ratio (intensity of NAD(P)H)/(intensity of FAD) was also calculated. FLIM data were combined with known metadata and analyzed to quantify the ability of metabolic imaging to differentiate embryos that resulted in pregnancy from embryos that did not. De-identified discarded aneuploid human embryos (n = 158) were also measured to quantify correlations with ploidy status and other factors. Statistical comparisons were performed using logistic regression and receiver operating characteristic (ROC) curves with 5-fold cross-validation averaged over 100 repeats with random sampling. AUC values were used to quantify the ability to distinguish between classes. No metabolic imaging parameters showed significant differences between good-quality blastocysts resulting in pregnancy versus those that did not. A logistic regression using metabolic data and metadata produced an ROC AUC of 0.58. In contrast, robust AUCs were obtained when classifying other factors such as comparison of Day 5 (n = 64) versus Day 6 (n = 41) blastocysts (AUC = 0.78), inner cell mass versus trophectoderm (n = 105: AUC = 0.88) and aneuploid (n = 158) versus euploid and positive pregnancy embryos (n = 108) (AUC = 0.82). The study protocol did not select which embryo to transfer and the cohort of 105 included blastocysts were all high quality. The study was also limited in number of participants and study sites. Increased power and performing the trial in more sites may have provided a stronger conclusion regarding the merits of the use of FLIM clinically. FLIM failed to distinguish consistent patterns in mitochondrial metabolism between good-quality blastocysts leading to pregnancy compared to those that did not. Blastocyst ploidy status was, however, highly distinguishable. In addition, embryo regions and embryo day were consistently revealed by FLIM. While metabolic imaging detects mitochondrial metabolic features in human blastocysts, this pilot trial indicates it does not have the potential to serve as an effective embryo viability detection tool. This may be because mitochondrial metabolism plays an alternative role post-implantation. This study was sponsored by Optiva Fertility, Inc. Boston IVF contributed to the clinical site and services. Becker Hickl, GmbH, provided the FLIM system on loan. T.S. was the founder and held stock in Optiva Fertility, Inc., and D.S. and E.S. had options with Optiva Fertility, Inc., during this study. The study was approved by WCG Connexus IRB (Study Number 1298156).

Comparative study of two models predicting the risk of deep vein thrombosis progression in spinal trauma patients after operation

ObjectivePatients with preoperative deep vein thrombosis (DVT) exhibit a notable incidence of postoperative deep vein thrombosis progression (DVTp), which bears a potential for silent, severe consequences. Consequently, the development of a predictive model for the risk of postoperative DVTp among spinal trauma patients is important. MethodsData of 161 spinal traumatic patients with preoperative DVT, who underwent spine surgery after admission, were collected from our hospital between January 2016 and December 2022. The least absolute shrinkage and selection operator (LASSO) combined with multivariable logistic regression analysis was applied to select variables for the development of the predictive logistic regression models. One logistic regression model was formulated simply with the Caprini risk score (Model A), while the other model incorporated not only the previously screened variables but also the age variable (Model B). The model’s capability was evaluated using sensitivity, specificity, positive predictive value, negative predictive value, accuracy, F1 score, and receiver operating characteristic (ROC) curve. Nomograms simplified and visually presented Model B for the clinicians and patients to understand the predictive model. The decision curve was used to analyze the clinical value of Model B. ResultsA total of 161 DVT patients were enrolled in this study. Postoperative DVTp occurred in 48 spinal trauma patients, accounting for 29.81% of the total patient enrolled. Model A inadequately predicted postoperative DVTp in spinal trauma patients, with ROC AUC values of 0.595 for the training dataset and 0.593 for the test dataset. Through the application of LASSO regression and multivariable logistic regression, a screening process was conducted for seven risk factors: D-dimer, blood platelet, hyperlipidemia, blood group, preoperative anticoagulant, spinal cord injury, lower extremity varicosities. Model B demonstrated superior and consistent predictive performance, with ROC AUC values of 0.809 for the training dataset and 0.773 for the test dataset. According to the calibration curves and decision curve analysis, Model B could accurately predict the probability of postoperative DVTp after spine surgery. The nomograms enhanced the interpretability of Model B in charts and graphs. ConclusionsIn summary, we established a logistic regression model for the accurate predicting of postoperative deep vein thrombosis progression in spinal trauma patients, utilizing D-dimer, blood platelet, hyperlipidemia, blood group, preoperative anticoagulant, spinal cord injury, lower extremity varicosities, and age as predictive factors. The proposed model outperformed a logistic regression model based simply on CRS. The proposed model has the potential to aid frontline clinicians and patients in identifying and intervening in postoperative DVTp among traumatic patients undergoing spinal surgery.

631. Using a commercially available assay that measures cytomegalovirus (CMV)-specific CD4+ and CD8+ T-cell immunity to predict protection against CMV: Results from a prospective, blinded clinical trial

Abstract Background Cytomegalovirus (CMV) infection is associated with significant morbidity and mortality among kidney transplant recipients (KTR). Given toxicities and costs of antivirals, there is growing interest in assays that measure CMV-specific T-cell immunity (TCI), which may predict protection against clinical infection. The Viracor® CMV TCI Panel (TCIP) uses flow cytometry and intracellular cytokine staining to measure %CMV-specific CD4+ and CD8+ T-cells; it is the only commercially available TCI assay in the US. In this investigator-initiated study, we evaluated the performance of the TCIP in predicting clinically significant CMV events. To our knowledge, this is the first prospective blinded clinical study of the TCIP. Methods We enrolled consecutive donor (D) or recipient (R) CMV seropositive KTR, who had TCIP testing monthly until either discontinuation of valganciclovir prophylaxis or the primary outcome. We also enrolled KTR with low-level untreated DNAemia, or after completion of valganciclovir treatment, to evaluate assay predictive values for progression or relapse of CMV infection, respectively. The primary outcome was CMV DNAemia (tested at the discretion of the treating provider for symptoms or laboratory findings), prompting treatment initiation. Secondary outcomes were any DNAemia and DNAemia &amp;gt;1000 IU/mL. Treating providers were unaware of TCIP results. Results We included 46 KTR. Median age was 50.5 (range 18-74) years. 73.6% identified as men, 10.8% Black, 8.7% Hispanic; 41.3% had CMV D+/R- status. KTR with CMV events had significantly lower %CMV-specific CD8+ T-cells (Fig. 1A, p=0.024), and the CMV protection ROC AUC was significant (Fig. 1B: AUC 0.78, 95%CI 0.55-0.99, p=0.026). The positive predictive values of both CD4+ and CD8+ T-cell positivity &amp;gt;0.2% for CMV protection were 96.3% (95%CI 81-99.9%) for the primary outcome, 92.6% (95%CI 75.7-99.1%) for any DNAemia, and 100% (one-sided 95%CI 87.2%) for DNAemia &amp;gt;1000 IU/mL.Figure 1.A. Comparison of CMV-specific %CD8+ T-cells between patients with or without (protected) CMV events. B Receiver Operating Characteristic (ROC) curve for CMV protection. Conclusion The TCIP exhibited good performance in predicting protection from CMV events and could be a useful adjunct tool in clinical decision-making, regarding initiation or extension of CMV prophylaxis or treatment, as well as adjustments in immunosuppression. Disclosures Panos Arvanitis, MS, NIH: Grant/Research Support|NIH: Brown University Summer Assistantship program and from the Brown Emerging Infectious Disease Scholars (EIDS) (5R25AI140490) Dimitrios Farmakiotis, M.D., Astellas: Grant/Research Support|Merck: Grant/Research Support|Viracor: Advisor/Consultant|Viracor: Grant/Research Support

Nomogram Based on Dual-Layer Spectral Detector CTA Parameter for the Prediction of Infarct Core in Patients with Acute Ischemic Stroke.

(1) Background: Acute ischemic stroke (AIS) is time-sensitive. The accurate identification of the infarct core and penumbra areas in AIS patients is an important basis for formulating treatment plans, and is the key to dual-layer spectral detector computed tomography angiography (DLCTA), a safer and more accurate diagnostic method for AIS that will replace computed tomography perfusion (CTP) in the future. Thus, this study aimed to investigate the value of DLCTA in differentiating infarct core from penumbra in patients with AIS to establish a nomogram combined with spectral computed tomography (CT) parameters for predicting the infarct core and performing multi-angle evaluation. (2) Methods: Data for 102 patients with AIS were retrospectively collected. All patients underwent DLCTA and CTP. The patients were divided into the non-infarct core group and the infarct core group, using CTP as the reference. Multivariate logistic regression analysis was used to screen predictors related to the infarct core and establish a nomogram model. The receiver operating characteristic (ROC) curve, the calibration curve, and decision curve analysis (DCA) were used to evaluate the predictive efficacy, accuracy, and clinical practicability of the model, respectively. (3) Results: Multivariate logistic analysis identified three independent predictors: iodine density (OR: 0.022, 95% CI: 0.003-0.170, p < 0.001), hypertension (OR: 7.179, 95% CI: 1.766-29.186, p = 0.006), and triglycerides (OR: 0.255, 95% CI: 0.109-0.594, p = 0.002). The AUC-ROC of the nomogram was 0.913. Calibration was good. Decision curve analysis was clinically useful. (4) Conclusions: The spectral CT parameters, specifically iodine density values, effectively differentiate between the infarct core and penumbra areas in patients with AIS. The nomogram, based on iodine density values, showed strong predictive power, discrimination, and clinical utility to accurately predict infarct core in AIS patients.

Accuracy of the GRACE Score for mortality prediction in myocardial infarction with nonobstructive coronary arteries (MINOCA)

Abstract Introduction Myocardial infarction with nonobstructive coronary arteries (MINOCA) patients account for 3.5 to 15% of all myocardial infarctions. The GRACE risk score is a well-validated tool for estimating short- and long-term risk in acute coronary syndrome, but the accuracy of the risk score for prognostic stratification in patients with MINOCA remains uncertain. Purpose We aimed to investigate whether the GRACE risk score can predict mortality in a real-world cohort of patients presenting with MINOCA. Methods We retrospectively analyzed patients admitted with MINOCA in a single center. Patients who were diagnosed with myocarditis and Takotsubo syndrome were excluded from the analysis. The GRACE risk score on admission was calculated. Patients were divided into a low-intermediate risk group (≤ 140 points, 71.5%) and a high-risk group (&amp;gt;140 points, 28.5%) according to their GRACE risk scores. Receiver operating characteristic (ROC) analysis was used to quantify the accuracy of the GRACE risk score to estimate both in-hospital and one-year mortality in MINOCA patients. Results Of a total of 3721 myocardial infarction patients, MINOCA was identified in 11.6% (n=430). 313 patients had complete data for the calculation of the GRACE on admission and were included in the analysis. Mean age was 65±12 years, and 49% (n=154) of patients were male. Mean GRACE score was 125±31 points. Patients in the high-risk group were older (74±9 vs. 61±11 years, p&amp;lt;0.001), had a higher incidence of in-hospital mortality (0.4% vs. 5.6%, p=0.008), and had a higher rate of one-year mortality (2.7% vs. 19.5%, p&amp;lt;0.001), compared with the low-intermediate risk group. There were no differences regarding sex and cardiovascular risk factors between the groups. In-hospital mortality was 1.9% and one-year mortality was 7.3%. ROC curve analysis showed that the GRACE risk score has a high value in predicting in-hospital mortality (AUC ROC 0.892; 95% CI 0.798–0.986; p=0.001), and one-year mortality (AUC ROC 0.826; 95% CI 0.739–0.0.914; p&amp;lt;0.001) in MINOCA patients. (Figure 1). The overall c-statistics were similar compared with estimates obtained from patients with myocardial infarction and significant coronary artery disease: in-hospital mortality (AUC ROC 0.0.849; 95% CI, 0.807–0.891; p&amp;lt;0.001); and one-year mortality (AUC ROC 0.776; 95% CI, 0.747–0.806; p&amp;lt;0.001). Conclusions In conclusion, the GRACE risk score provides potentially valuable prognostic information on in-hospital and one-year mortality when applied to MINOCA patients.Figure 1

Stress-only dynamic computed tomography perfusion protocol (CTP) alone without computed tomography coronary angiography (CCTA) has limited specificity to diagnose ischemia: A retrospective two-center study

PurposeTo investigate diagnostic performance of stress-only dynamic myocardial computed tomography perfusion (CTP) without computed tomography coronary angiography (CCTA) to diagnose ischemia with invasive fractional flow reserve (FFR) as a reference standard. Method135 datasets (68 positive for ischemia with invasive FFR < 0.8) acquired with a 256-slice CT system (Revolution, GE Healthcare, Chicago, IL, USA) were retrieved, postprocessed with a deep learning-based algorithm (Advanced intelligent Clear-IQ Engine (AiCE), Canon Medical Systems, Otawara, Japan) (FC03/cardiac kernel, 8 mm slice thickness), analyzed using a dedicated workstation (Vitrea research 7.11.0. Vital Images, Minnetonka, MN, USA), and loaded into a clinical workstation (CardIQ, GE Healthcare, Chicago, IL, USA) for review. Ten observers with various experience from two research sites evaluated the post-processed images, perfusion slices and maps to indicate presence vs absence of perfusion defect and its probability (five-point Likert scale). Binary decisions and probability scores were used to calculate sensitivity and specificity for each reader, and to create receiver operating characteristics (ROC) curves, respectively. Furthermore, the correlation coefficient (ICC) was computed. ROC AUC of a purely quantitative analysis was obtained thanks to a color-coded map with a fixed scale superimposed on myocardial walls displaying myocardial blood flow (MBF) values. ResultsThe overall case-based sensitivity and specificity for the detection of perfusion deficit were 0.79 and 0.30, respectively. No significant differences were detected in the AUC across readers (p value = 0.66). The AUC values were 0.50, 0.58, 0.63, 0.59, 0.45, 0.60, 0.56, 0.61, 0.52, 0.61. Absolute reader agreement ICC was 0.60 (good agreement) for an average case. ConclusionDynamic CTP alone has good sensitivity, but low specificity when analyzed without CCTA. These findings reinforce the need to guide the interpretation functional test with the knowledge of coronary artery anatomy.

UTILITY OF STREM-1 BIOMARKER AND HCP GENE FOR IDENTIFICATION OF ACINETOBACTER BAUMANNII COLONIZATION AND INFECTION IN LUNG.

Objective: Respiratory infections or colonization of Acinetobacter baumannii (Ab) are common in clinical practice but are treated differently. Early identification of Ab infection and colonization reduces the risk of antibiotic mismatch but objective laboratory indicators to distinguish between bacterial infections and colonization are lacking. To distinguish infection and colonization of Ab, we tested the role of two biomarkers, triggering receptor expressed on myeloid cells-1 (TREM-1) and hemolysin coregulated protein. Methods: A total of 96 inpatients with Ab were divided into infection and colonization groups. Blood samples were collected on days 1, 2, 3, 5, 8, and 10 and daily maximum body temperature was recorded. Polymerase Chain Reaction and Reverse Transcription Polymerase Chain Reaction were used to detect the presence and expression levels of the hcp gene in Ab clinical isolates. Results : sTREM-1 and procalcitonin (PCT) levels on days 1 to 10 and neutrophil classification (N%) on days 1 to 3 were different ( P < 0.05) in the infection group and colonization group. Receiver operating characteristic (ROC) curves showed significant differences in N% and sTREM-1 on days 2 and 3 ( P < 0.01). sTREM-1 had the highest AUC ROC on days 1, 2, and 3 of all the markers. On day 1, the ROC curve of "WBC&N%&PCT&sTREM-1" was statistically different from individual indices (white blood cell count, N%, and PCT; P < 0.05) and was equal to the ROC curve of sTREM-1 ( P > 0.05). Thirty five of 96 patients were classified as infection group and 61 as colonization group with hcp gene detection rates of 71.43% (25/35) and 31.15% (19/61), respectively. No differences in hcp gene presence and transcript levels were found between two groups ( P > 0.05). Conclusions: Dynamic monitoring of sTREM-1 and PCT is valuable in identifying Ab infection and colonization. sTREM-1 can be improved by combination with multiple biomarkers in the early stage for identification of infection and colonization. The hcp gene was more likely to be present in the infection cohort.

New Score Models for Assessing Disease Activity in Crohn's Disease Based on Bowel Ultrasound and Biomarkers: Ideal Surrogates for Endoscopy or Imaging.

Disease activity evaluation is important in Crohn's disease (CD). We aimed to establish new disease activity indices for CD based on noninvasive parameters. The data of 110 CD patients were retrospectively analyzed. Parameters from bowel ultrasound and biomarkers were measured to select the variables included in the models by univariate analysis. Logistic regression analysis was performed to predict mucosal and transmural activities defined by ileocolonoscopy or computed tomography enterography (CTE), respectively. The models' performance was measured by the area under the receiver operating characteristic (ROC) curve (AUC). Leave-one-out cross validation (LOOCV) was applied to adjust for overconfidence in the newly established score models. To predict mucosal activity, erythrocyte sedimentation rate (ESR) and (LimG × BWT)-SUM (the sum of the product of Limberg grade [LimG] and bowel wall thickness [BWT] of each bowel segment) were selected for Model A, and the equation was A=2 × ESR + 9.3 × (LimG × BWT)-SUM. The AUC of ROC, sensitivity, and specificity were 0.927, 89.8%, and 86.4%, respectively. The AUC of the ROC curve verified by LOOCV was 0.913. To predict transmural activity, albumin (ALB) and LimG-SUM (the sum of the LimG of all the bowel segments) were selected for Model B, which was established as B=-1.3 × ALB + 1.7 × LimG-SUM. The AUC of ROC, sensitivity, and specificity were 0.851, 78.0%, and 84.2%, respectively. The AUC of the ROC curve verified by LOOCV was 0.833. Nomograms were developed for two score models. New score models based on noninvasive parameters established in this study showed good abilities in detecting active disease and performed well in the validation phase.

#3802 ROLE OF NEUTROPHIL EXTRACELLULAR TRAPS IN DIABETIC KIDNEY DISEASE: DATA FROM INTEGRATION OF BULK RNA AND SINGLE-NUCLEUS RNA SEQUENCING ANALYSES

Abstract Background and Aims Neutrophil extracellular traps (NETs) is an important immune response against infections that is implicated in various immune-mediated conditions. Growing evidence shows that inflammation has key pathogenic contributions in diabetic kidney disease (DKD) but the role of NETs remains unclear. Method We evaluated the differentially expressed genes (DEGs) of NETs in human DKD using the bulk RNA sequencing datasets of kidney biopsy from DKD patients (GSE30528 and GSE30529). The candidate genes were further selected based on the machine learning algorithms (LASSO and SVM-RFE). The single-nucleus RNA sequencing (GSE195460 and GSE131882) and DKD bulk RNA sequencing (GSE30122) were used to validate the results. Receiver Operating Characteristics (ROC) curve were constructed to assess the diagnostic performance of the identified genes for DKD. Candidate drugs were further screened from the DSigDB database. Results Three characteristic genes (ITGAM, ITGB2 and TLR7) were selected using machine learning approach, which were all upregulated in human DKD glomerular and tubulointerstitial portion compared to healthy controls (all P &amp;lt;0.05). Single-cell analysis further demonstrated that ITGAM, ITGB2 and TLR7 were mainly overexpressed in leucocytes. ITGAM, ITGB2 and TLR7 expression in tubulointerstitial compartment also showed good diagnostic performance for DKD (ROC AUC 0.983, 0.975 and 0.958 respectively). Candidate drugs that target ITGAM, ITGB2 or TLR7 genes in DKD include ropivacaine, lidocaine and sulfasalazine. Conclusion ITGAM, ITGB2 and TLR7 may serve as biomarkers for DKD and drugs that target these genes may have therapeutic potential for DKD.

Assessment of Brain Computed Tomography as a Practical Way to Predict Neurological Outcomes in Post-Cardiac Arrest Survivors

Objective: To evaluate whether a visual inspection of ischemic changes in brain computed tomography (CT) images, which is very practical in clinical settings, could predict neurological outcomes in post-cardiac arrest survivors. Materials and Methods: The authors retrospectively reviewed medical records and brain CT images of 62 patients who remained comatose after surviving cardiac arrest and had undergone a CT scan within 24 hours to seven days. Neurological outcomes at one month were assessed using the Cerebral Performance Category (CPC). The CPC scores of 1-2 and 3-5 referred to good and poor neurological outcomes, respectively. Findings from the inspection of the images were graded into grade 1 for absence of acute ischemic change, grade 2 for decreased attenuation of the grey matter in some brain region(s), and grade 3 for diffused loss of grey-white matter differentiation or apparent brain swelling. An experienced neurologist and a consensus group of four pre-clerkship medical students, blinded to the neurological outcomes, evaluated the grade of ischemic changes in CT images. Results: Positive correlations were observed between CPC and CT grading by both the neurologist (ρ=0.76, 95% CI 0.63 to 0.90, p&lt;0.001) and medical students (ρ=0.57, 95% CI 0.38 to 0.77, p&lt;0.001). The CT grading of 2 or more by the neurologist could predict poor neurological outcomes with specificity of 1.00, sensitivity of 0.89, and receiver operating characteristic (ROC) AUC of 0.94 (95% CI 0.89 to 1.00). The evaluation by medical students showed an ROC AUC of 0.80 (95% CI 0.64 to 0.96). Conclusion: The simple visual inspection of ischemic changes in brain CT images showed a high diagnostic accuracy and could be a practical method for predicting neurological outcomes in post-cardiac arrest survivors. Keywords: Cardiac arrest; Post-cardiac arrest ischemic brain injury; Neurological outcomes; Computerized tomography

The predictive value of warning scores for intensive care unit admission in coronavirus disease 2019 patients

To evaluate the predictive value of early warning scores for intensive care unit (ICU) admission in patients with coronavirus disease 2019 (COVID-19). For COVID-19 patients who were admitted to Shijiazhuang People's Hospital from January 2021 to February 2021, national early warning score (NEWS), national early warning score 2 (NEWS2), rapid emergency medicine score (REMS), quick sepsis-related organ failure (qSOFA), altered consciousness, blood urea nitrogen, respiratory rate, blood pressure, and age-65 (CURB-65) were used to evaluate the inpatient condition and the predictive value for ICU admission. A total of 368 patients were included, and 32 patients (8.7%) were transferred to the ICU. The median age was 49.0 (34.0,61.0) years. The scores of NEWS, NEWS2, REMS, and CURB-65 were 1 (0, 2), 1 (0, 2), 4 (2, 6) and 0 (0, 1), respectively. The receiver operating characteristic (ROC) cure (AUC) was used to evaluate the predictive value in detecting patients who are at risk of being transferred to the ICU. Area under the ROC AUC of NEWS was 0.756, sensitivity 65.6%, and specificity 71.3%. ROC AUC of NEWS2 was 0.732, sensitivity 62.5%, and specificity 61.3%. ROC AUC of REMS was 0.787, sensitivity 84.4%, and specificity 64.6%. ROC AUC of CURB-65 was 0.814, sensitivity 81.3%, and specificity 76.8%. The predictive value of NEWS and NEWS2 combined with age were significantly improved. The ROC AUC of NEWS combined with age was 0.885, sensitivity 85.1%, and specificity 75.0%. The ROC AUC of NEWS2 combined with age was 0.883, sensitivity 84.2%, and specificity 75.0%. NEWS and NEWS2 combined with age can be used as a predictive tool for whether COVID-19 patients will be admitted to the ICU.

Identification of HOXB9 to predict prognosis of endometrial cancer based on comprehensive bioinformatics analysis

BackgroundThe HOXB9 gene, which plays a key role in embryonic development, is also involved in the regulation of various human cancers. However, the potential relationship between HOXB9 and endometrial cancer (EC) has not yet been comprehensively analyzed and fully understood.MethodsWe used multiple bioinformatics tools to explore the role of HOXB9 in EC.ResultsThe expression of HOXB9 was significantly upregulated in pan-cancer, including EC (P < 0.05). Quantitative real time polymerase chain reaction (qRT-PCR) experiment confirmed the high expression of HOXB9 in EC from clinical samples (P < 0.001). Double validated by Enrichr and Metascape, HOXB9 showed a strong correlation with HOX family, suggesting that HOX family may also involve in the development of EC (P < 0.05). Enrichment analysis revealed HOXB9 is mainly associated with cellular process, developmental process, P53 signaling pathway, etc. At the single-cell level, the clusters of cells ranked were glandular and luminal cells c-24, glandular and luminal cells c-9, endothelial cells c-15, compared with the other cells. At the genetic level, promoter methylation levels of HOXB9 were significantly higher in tumors than in normal tissues. Furthermore, variations of HOXB9 were closely associated with overall survival (OS) and recurrence free survival (RFS) in EC patients (P < 0.05). The agreement between univariate and multivariate Cox regression indicated that the results were more reliable. Stages III and IV, G2 and G3, tumor invasion ≥ 50%, mixed or serous histological type, age > 60 years, and high expression of HOXB9 were risk factors strongly associated with OS in EC patients (P < 0.05). Therefore, six factors were incorporated to construct a nomogram for survival prediction. Finally, we used the Kaplan-Meier (KM) curve, receiver operating characteristic (ROC) curve, and time-dependent ROC to assess predictive power of HOXB9. KM curve showed EC patients overexpressing HOXB9 had a worse OS. AUC of diagnostic ROC was 0.880. AUCs of time-dependent ROC were 0.602, 0.591, and 0.706 for 1-year, 5-year, and 10-year survival probabilities (P < 0.001).ConclusionsOur study provids new insights into the diagnosis and prognosis of HOXB9 in EC and constructs a model that can accurately predict the prognosis of EC.

Estimation of the ROC curve and the area under it with complex survey data

Logistic regression models are widely applied in daily practice. Hence, it is necessary to ensure they have an adequate predictive performance, which is usually estimated by means of the receiver operating characteristic (ROC) curve and the area under it (area under the curve [AUC]). Traditional estimators of these parameters are thought to be applied to simple random samples but are not appropriate for complex survey data. The goal of this work is to propose new weighted estimators for the ROC curve and AUC based on sampling weights which, in the context of complex survey data, indicate the number of units that each sampled observation represents in the population. The behaviour of the proposed estimators is evaluated and compared with the traditional unweighted ones by means of a simulation study. Finally, weighted and unweighted ROC curve and AUC estimators are applied to real survey data in order to compare the estimates in a real scenario. The results suggest the use of the weighted estimators proposed in this work in order to obtain unbiassed estimates for the ROC curve and AUC of logistic regression models fitted to complex survey data.

A Prediction Model for Osteoporosis Risk Using a Machine-Learning Approach and Its Validation in a Large Cohort.

Osteoporosis develops in the elderly due to decreased bone mineral density (BMD), potentially increasing bone fracture risk. However, the BMD is not regularly measured in a clinical setting. This study aimed to develop a good prediction model for the osteoporosis risk using a machine learning (ML) approach in adults over 40 years in the Ansan/Anseong cohort and the association of predicted osteoporosis risk with a fracture in the Health Examinees (HEXA) cohort. The 109 demographic, anthropometric, biochemical, genetic, nutrient, and lifestyle variables of 8,842 participants were manually selected in an Ansan/Anseong cohort and included in the ML algorithm. The polygenic risk score (PRS) of osteoporosis was generated with a genome-wide association study and added for the genetic impact of osteoporosis. Osteoporosis was defined with < -2.5 T scores of the tibia or radius compared to people in their 20s-30s. They were divided randomly into the training (n = 7,074) and test (n = 1,768) sets-Pearson's correlation between the predicted osteoporosis risk and fracture in the HEXA cohort. XGBoost, deep neural network, and random forest generated the prediction model with a high area under the curve (AUC, 0.86) of the receiver operating characteristic (ROC) with 10, 15, and 20 features; the prediction model by XGBoost had the highest AUC of ROC, high accuracy and k-fold values (> 0.85) in 15 features among seven ML approaches. The model included the genetic factor, genders, number of children and breastfed children, age, residence area, education, seasons to measure, height, smoking status, hormone replacement therapy, serum albumin, hip circumferences, vitamin B6 intake, and body weight. The prediction models for women alone were similar to those for both genders, with lower accuracy. When the prediction model was applied to the HEXA study, the correlation between the fracture incidence and predicted osteoporosis risk was significant but weak (r = 0.173, P < 0.001). The prediction model for osteoporosis risk generated by XGBoost can be applied to estimate osteoporosis risk. The biomarkers can be considered for enhancing the prevention, detection, and early therapy of osteoporosis risk in Asians.