Receiver Operator Characteristic Plots Research Articles

Serum Levels of Arachidonic Acid, Interleukin-6, and C-Reactive Protein as Potential Indicators of Pulmonary Viral Infections: Comparative Analysis of Influenza A, Respiratory Syncytial Virus Infection, and COVID-19

Acute respiratory tract infections, including influenza A (FluA), respiratory syncytial virus (RSV) infection, and COVID-19, can aggravate to levels requiring hospitalization, increasing morbidity and mortality. Identifying biomarkers for an accurate diagnosis and prognosis of these infections is a clinical need. We performed a cross-sectional study aimed to investigate the changes in circulating levels of arachidonic acid, interleukin 6 (IL-6), and C-reactive protein (CRP) in patients with FluA, RSV, or COVID-19, and to analyze the potential of these parameters as diagnosis or prognosis biomarkers. We analyzed serum samples from 172 FluA, 80 RSV, and 217 COVID-19 patients, and 104 healthy volunteers. Individuals with lung viral diseases showed reduced arachidonic acid concentrations compared to healthy people, with these differences being most pronounced in the order COVID-19 > RSV > FluA. Conversely, IL-6 and CRP levels were elevated across diseases, with IL-6 emerging as the most promising diagnostic biomarker, with areas under the curve (AUC) of the receiver operating characteristics plot higher than 0.85 and surpassing arachidonic acid and CRP. Moreover, IL-6 displayed notable efficacy in distinguishing between FluA patients who survived and those who did not (AUC = 0.80). These findings may provide useful tools for diagnosing and monitoring the severity of acute viral respiratory tract infections, ultimately improving patient outcomes.

External validation and updating of the Rossi nomogram for predicting cesarean delivery following induction: is the Bishop score valuable?

This study sought to validate the Rossi nomogram in a Chinese population and then to include the Bishop score to see if it has an effect on the accuracy of the nomogram. The Rossi predictive model was applied and externally validated in a retrospective cohort from August 2017 and July 2023 in a Chinese tertiary-level medical center. For the revision and updating of the models, the regression coefficients of all the predictors (except race) were re-estimated and then the cervical Bishop score at the time of induction was added. Each model's performance was measured using the receiver-operating characteristic and calibration plots. Decision curve analysis determined the range of the probability threshold for each prediction model that would be of clinical value. A total of 721 women met the inclusion criteria, of whom 183 (25.4%) underwent a cesarean delivery. The calibration demonstrated the underestimation of the original model, with an area under the curve (AUC) of 0.789 (95% confidence interval [CI] 0.753-0.825, p < 0.001). After recalibrating the original model, the discriminative performance was improved from 0.789 to 0.803. Moreover, the discriminatory power of the updated model was further improved when the Bishop score at the time of induction was added to the recalibrated multivariable model. Indeed, the updated model demonstrated good calibration and discriminatory power, with an AUC of 0.811. The decision curve analysis indicated that all the models (original, recalibrated, and updated) provided higher net benefits of between 0 and 60% of the probability threshold, which indicates the benefits of using the models to make decisions concerning patients who fall within the identified range of the probability threshold. The net benefits of the updated model were higher than those of the original model and the recalibrated model. The nomogram used to predict cesarean delivery following induction developed by Rossi et al. has been validated in a Chinese population in this study. More specifically, adaptation to a Chinese population by excluding ethnicity and including the Bishop score prior to induction gave rise to better performance. The three models (original, recalibrated, and updated) offer higher net benefits when the probability threshold is between 0 and 60%.

Analysis and validation of biomarkers of immune cell-related genes in postmenopausal osteoporosis: An observational study.

Postmenopausal osteoporosis (PMOP) is a common metabolic inflammatory disease. In conditions of estrogen deficiency, chronic activation of the immune system leads to a hypo-inflammatory phenotype and alterations in its cytokine and immune cell profile, although immune cells play an important role in the pathology of osteoporosis, studies on this have been rare. Therefore, it is important to investigate the role of immune cell-related genes in PMOP. PMOP-related datasets were downloaded from the Gene Expression Omnibus database. Immune cells scores between high bone mineral density (BMD) and low BMD samples were assessed based on the single sample gene set enrichment analysis method. Subsequently, weighted gene co-expression network analysis was performed to identify modules highly associated with immune cells and obtain module genes. Differential analysis between high BMD and low BMD was also performed to obtain differentially expressed genes. Module genes are intersected with differentially expressed genes to obtain candidate genes, and functional enrichment analysis was performed. Machine learning methods were used to filter out the signature genes. The receiver operating characteristic (ROC) curves of the signature genes and the nomogram were plotted to determine whether the signature genes can be used as a molecular marker. Gene set enrichment analysis was also performed to explore the potential mechanism of the signature genes. Finally, RNA expression of signature genes was validated in blood samples from PMOP patients and normal control by real-time quantitative polymerase chain reaction. Our study of PMOP patients identified differences in immune cells (activated dendritic cell, CD56 bright natural killer cell, Central memory CD4 T cell, Effector memory CD4 T cell, Mast cell, Natural killer T cell, T follicular helper cell, Type 1 T-helper cell, and Type 17 T-helper cell) between high and low BMD patients. We obtained a total of 73 candidate genes based on modular genes and differential genes, and obtained 5 signature genes by least absolute shrinkage and selection operator and random forest model screening. ROC, principal component analysis, and t-distributed stochastic neighbor embedding down scaling analysis revealed that the 5 signature genes had good discriminatory ability between high and low BMD samples. A logistic regression model was constructed based on 5 signature genes, and both ROC and column line plots indicated that the model accuracy and applicability were good. Five signature genes were found to be associated with proteasome, mitochondria, and lysosome by gene set enrichment analysis. The real-time quantitative polymerase chain reaction results showed that the expression of the signature genes was significantly different between the 2 groups. HIST1H2AG, PYGM, NCKAP1, POMP, and LYPLA1 might play key roles in PMOP and be served as the biomarkers of PMOP.

Assessment of balance and fear of falling in individuals with acute ankle sprain: A cross-sectional pilot study

Objectives: The primary objective of this research was to assess balance, pain, and fear of falling (FOF) among individuals with an acute ankle sprain. The secondary objective was to correlate FOF with pain and balance. Methods: The research was conducted among individuals with acute ankle sprain —two to three days after removal of two weeks of casting at M S Ramaiah hospitals, Bengaluru. The primary outcome measures were the Short Falls Efficacy Scale-International (Short FES-I) and the star excursion balance test (SEBT), and the secondary outcome measure was the numerical pain rating scale (NPRS). Cutoff points for Short FES-I defined in prior research using receiver operator characteristic plots were used in this study. For Short FES-I, frequencies and percentages were employed. The NPRS and SEBT were measured using means and standard deviations. Results: The results showed that among 21 individuals, 76.2% had FOF. Among those with FOF, 56.25% had moderate FOF, and 43.75% had severe FOF. Mean SEBT scores were greater for the uninvolved limb than for the involved limb. Spearman’s Rho test showed that the Short FES-I and NPRS showed no correlation (P = 0.102 &gt; 0.05), and Short FES-I and SEBT showed a significant correlation (P = 0.003 &lt; 0.05). Conclusion: The FOF was present in two-thirds of those with acute ankle sprains, and balance impairment was present in individuals with Grade II ankle sprains. Among individuals with FOF, the degree of FOF was moderate to severe. There was a significant negative correlation between FOF and balance, but no correlation was found between FOF and pain.

An AI-based multiphase framework for improving the mechanical ventilation availability in emergency departments during respiratory disease seasons: a case study

BackgroundShortages of mechanical ventilation have become a constant problem in Emergency Departments (EDs), thereby affecting the timely deployment of medical interventions that counteract the severe health complications experienced during respiratory disease seasons. It is then necessary to count on agile and robust methodological approaches predicting the expected demand loads to EDs while supporting the timely allocation of ventilators. In this paper, we propose an integration of Artificial Intelligence (AI) and Discrete-event Simulation (DES) to design effective interventions ensuring the high availability of ventilators for patients needing these devices.MethodsFirst, we applied Random Forest (RF) to estimate the mechanical ventilation probability of respiratory-affected patients entering the emergency wards. Second, we introduced the RF predictions into a DES model to diagnose the response of EDs in terms of mechanical ventilator availability. Lately, we pretested two different interventions suggested by decision-makers to address the scarcity of this resource. A case study in a European hospital group was used to validate the proposed methodology.ResultsThe number of patients in the training cohort was 734, while the test group comprised 315. The sensitivity of the AI model was 93.08% (95% confidence interval, [88.46 − 96.26%]), whilst the specificity was 85.45% [77.45 − 91.45%]. On the other hand, the positive and negative predictive values were 91.62% (86.75 − 95.13%) and 87.85% (80.12 − 93.36%). Also, the Receiver Operator Characteristic (ROC) curve plot was 95.00% (89.25 − 100%). Finally, the median waiting time for mechanical ventilation was decreased by 17.48% after implementing a new resource capacity strategy.ConclusionsCombining AI and DES helps healthcare decision-makers to elucidate interventions shortening the waiting times for mechanical ventilators in EDs during respiratory disease epidemics and pandemics.

Comparing 2-dimensional macular pigment optical density with objective and subjective perimetry and visual acuity in age-related macular degeneration.

To compare diagnostic power for different severities of age-related macular degeneration (AMD) of two-dimensional macular pigment optical densities (2D-MPOD) and spatially matched objective perimetry, with standard perimetry and best-corrected visual acuity (BCVA). The ObjectiveField Analyser (OFA) provided objective perimetry, and a Heidelberg Spectralis optical coherence tomography (OCT) measured 2D-MPOD in AMD patients, both completed twice over 0.99 ± 0.16years. From each 2D-MPOD image, we extracted 20 regions/macula, matched to the 20 OFA stimuli/macula. For each region, we calculated 7 measures from the 2D-MPOD pixel values and correlated those with OFA sensitivities and delays. We quantified 2D-MPOD changes, the ability of 2D-MPOD and OFA to discriminate AMD stages, and the discriminatory power of Matrix perimetry and BCVA using percentage area under receiver operator characteristic plots (%AUROC). In 58 eyes of 29 subjects (71.6 ± 6.3years, 22 females), we found significant correlations between 2D-MPOD and OFA sensitivities for Age-Related Eye Disease Studies (AREDS)-3 and AREDS-4 severities. Delays showed significant correlations with AREDS-2. For AREDS-4, correlations extended across all eccentricities. Regression associated with the Bland-Altman plots showed significant changes in 2D-MPOD over the study period, especially variability measures. MPOD per-region medians discriminated AREDS-1 from AREDS-3 eyes at a %AUROC of 80.0 ± 6.3%, outperforming OFA, Matrix perimetry, and BCVA. MPOD changes correlated with central functional changes and significant correlations extended peripherally in later-stage AMD. Good diagnostic power for earlier-stage AMD and significant change over the study suggest that 2D-MPOD and OFA may provide effective biomarkers.

Performance evaluation and verification of post-processing methods for TIGGE ensemble data using machine learning approaches

Abstract Ensemble modelling has become a significant technique in the field of machine learning, as it utilises the combined knowledge of multiple base models to improve the accuracy of predictions in different domains. Nevertheless, the effectiveness of ensemble predictions relies on the implementation of post-processing techniques that enhance and optimize the outputs of the ensemble. This study explores the domain of ensemble data post-processing, utilizing a machine learning-focused methodology to thoroughly assess and contrast a variety of post-processing methods. TIGGE Ensemble data from ECMWF and NCEP were used from 2010 to 2020. Research covers machine learning approaches post-processing methods such as BMA, cNLR, HXLR, OLR, logreg, hlogreg, QM were applied. The probabilistic forecasts were validated using the Brier Score (BS), Area Under Curve (AUC) of Receiver Operator Characteristics (ROC) plots and reliability plots. The cNLR and BMA strategies for post-processing performed exceptionally well with BS value of 0.10 and RPS value of 0.11 at all grid points for both methods. The ROC–AUC values for the cNLR and BMA methods were found to be 91.87 and 91.82%, respectively. The results show that improved post-processing techniques can be helpful to predict the flood in advance with accurate precision and warning.

Levels of Angiotensin and Kinin Metabolite Peptides Related to COVID-19 Severity.

In addition to crucial roles in normal human biology, peptide metabolites of the renin-angiotensin (RAS) and kallikrein-kinin systems (KKS) have been reported to be altered in COVID-19 patients. Here, we evaluate new data on RAS and KKS peptides in COVID-19 patient serum obtained from a recently developed, fully validated, and optimized stable isotope labeling LC-MS peptide assay. We found that the RAS peptides angiotensin (ANG) 1, 2, 1-5, and 1-7 were downregulated compared to COVID-free surrogate controls, while the KKS peptides Brad, Brad 1-8, and Brad 1-7 were upregulated. This paper focuses on uncovering the possible diagnostic value of these peptides using receiver operating characteristic (ROC) analyses of these data. ROC plots confirmed that all of the analyte peptides in 80 serum samples from COVID-19 patients were significantly altered from "normal" values of the control samples. The best diagnostic sensitivities and selectivities for COVID vs no COVID were found in ROC plots for Brad and Brad 1-7 (both 99% sensitivity, 100% selectivity). We then analyzed levels of all the peptides grouped according to preassigned values of the World Health Organization (WHO) COVID-19 Severity Index. ROC plots differentiated patients with a high WHO severity index from those with a low WHO severity index with moderate success, with BRAD (73% sensitivity, 79% selectivity) and Ang 1-7 (75% sensitivity, 65% selectivity) giving the best diagnostic performance. Results suggest the possible diagnostic value of these peptides as biomarkers to help identify moderate and serious COVID-19 cases at relatively early stages.

The Role of C-reactive Protein and Procalcitonin in Predicting the Occurrence of Pancreatic Fistula in Patients who Underwent Laparoscopic Pancreaticoduodenectomy: a Retrospective Study.

The occurrence of postoperative pancreatic fistula following laparoscopic pancreaticoduodenectomy (LPD) is a significant concern, yet there is currently a lack of consensus on reliable predictive methods for this complication. Therefore, the aim of this study was to assess the clinical significance of C-reactive protein (CRP) and procalcitonin (PCT) values and their reliability in early predicting the development of clinically relevant pancreatic fistula (CRPF) following LPD.A retrospective analysis was conducted using data from 120 patients who had LPD between September 2019 and December 2021. Preoperative assessment data, standard patients' demographic and clinicopathological characteristics, intra- and postoperative evaluation, as well as postoperative laboratory values on postoperative days (PODs) 1, 3, and 7, including white blood cells (WBCs), CRP, and PCT, were prospectively recorded on a dedicated database. Two clinicians separately collected and cross-checked all of the data.Among 120 patients [77 men (64%), 43 women (36%], CRPF occurred in 15 patients (11 grade B and 4 grade C fistulas). The incidence rate of CRPF was 12.3%. A comparison of the median values of WBCs, PCT, and CRP across the two groups revealed that the CRPF group had higher values on most PODs than the non-CRPF group. Receiver operating characteristic (ROC) analysis was used to calculate the area under the curve (AUC) and cutoff values. It was discovered that POD 3 has the most accurate and significant values for WBCs, CRP, and PCT. According to the ROC plots, the AUC for WBCs was 0.842, whereas the AUC for PCT was 0.909. As for CRP, the AUC was 0.941 (95% CI 0.899-0.983, p<0.01) with a cutoff value of 203.45, indicating a sensitivity of 93.3% and specificity of 91.4%.Both CRP and PCT can be used to predict the early onset of CRPF following LPD, with CRP being slightly superior on POD 3.

The predictive value of a computed tomography-based radiomics model for the surgical separability of thymic epithelial tumors from the superior vena cava and the left innominate vein.

The aim of this study was to develop a radiomics machine learning model based on computed tomography (CT) that can predict whether thymic epithelial tumors (TETs) can be separated from veins during surgery and to compare the accuracy of the radiomics model to that of radiologists. Patients who underwent thymectomy at our hospital from 2009 to 2017 were included in the screening process. After the selection of patients according to the inclusion and exclusion criteria, the cohort was randomly divided into training and testing groups, and CT images of these patients were collected. Subsequently, two-dimensional (2D) and three-dimensional (3D) regions of interest were labelled using ITK-SNAP 3.8.0 software, and Radiomics features were extracted using Python software (Python Software Foundation) and selected through the least absolute shrinkage and selection operator (LASSO) regression model. To construct the classifier, a support vector machine (SVM) was employed, and a nomogram was created using logistic regression to predict vascular inseparable TETs based on the radiomics score (radscore) and image features. To assess the accuracy of these models, area under receiver operating characteristic (ROC) curves of these models were calculated, and differences among the models were identified using the Delong test. In this retrospective study, 204 patients with TETs were included, among whom 21 were diagnosed with surgical vascularly inseparable TETs. The area under ROC curve (AUC) of the 2D model, 3D model, 2D + 3D model, and radiologist diagnoses were 0.94, 0.92, 0.95, and 0.87 in the training cohort and 0.95, 0.92, 0.98, and 0.78 in testing cohort, respectively. The Delong test revealed a significant improvement in the performance of the radiomics models compared to radiologists' diagnoses. The logistic regression selected 3 image features, namely maximum diameter of the tumor, degree of abutment of vessel circumference >50%, and absence of the mediastinal fat layer or space between the tumor and surrounding structures. These features, along with the radscore, were included to develop a nomogram. The AUCs of this nomogram were 0.99 in both the training set and testing set, and the Delong test did not find a significant difference between ROC plots of the nomogram and radiomics models. The proposed radiomics model could accurately predict surgical vascularly inseparable TETs preoperatively and was shown to have a higher predictive value than the radiologists.

Geospatial modelling of mineral potential zones using data-driven based weighting factor and statistical index techniques

Mineral prospectivity models (MPMs) are significantly essential in delineating target zones with the optimum likelihood of containing a particular sought-after mineral deposit. This present study carried out mineral potential mapping over the Collette Prospecting Licence (PL) Area of north-western Ghana using bivariate data-driven spatial statistical models composed of statistical index (SI) and weighting factor (WF) approaches. In the first instance, the geographic coordinates of variously known locations of artisanal mining operations as well as high Au concentration locations were mapped during a field survey. As a result, 181 known locations of Au occurrences were identified, out of which 127 (70%) were selected randomly for training and creating the mineral prospectivity models, whereas the remaining 54 (30%) were used to assess and validate the accuracy of the predictive models produced. The efficacy of mineral prospectivity models generated enormously depends on the appropriate selection of mineral-related factors. In this study, the following mineral-related condition factors (evidential layers) comprising analytic signal, lineament density, uranium-thorium ratio, uranium, potassium-thorium ratio, potassium, reduction-to-equator, and geology were used. The aforementioned evidential layers were derived and sourced from geophysical and geological datasets, which were later prepared for the generation of the models in a geographic information systems (GIS) environment. Finally, the validation of the mineral prospectivity models generated was carried out by applying the receiver operating characteristics (ROC) curve. The estimated results based on the ROC plots obtained for the predictive models showed that the area under the ROC curve (AUC) scores obtained for the SI-based and WF-based mineral prospectivity models were respectively, 0.780 and 0.733. Hence, it can be concluded that both mineral predictive models created in this study produced reasonably good accuracy (AUC score greater than 0.7) in predicting the potential zones of gold mineralisation occurrences within the Collette PL Area of north-western Ghana. These MPMs can serve as essential models for mineral exploration programmes within the study area.

Recognition of Pediatric Congenital Heart Diseases by Using Phonocardiogram Signals and Transformer-Based Neural Networks.

The phonocardiogram (PCG) or heart sound auscultation is a low-cost and non-invasive method to diagnose Congenital Heart Disease (CHD). However, recognizing CHD in the pediatric population based on heart sounds is difficult because it requires high medical training and skills. Also, the dependency of PCG signal quality on sensor location and developing heart in children are challenging. This study proposed a deep learning model that classifies unprocessed or raw PCG signals to diagnose CHD using a one-dimensional Convolution Neural Network (1D-CNN) with an attention transformer. The model was built on the raw PCG data of 484 patients. The results showed that the attention transformer model had a good balance of accuracy of 0.923, a sensitivity of 0.973, and a specificity of 0.833. The Receiver Operating Characteristic (ROC) plot generated an Area Under Curve (AUC) value of 0.964, and the F1-score was 0.939. The suggested model could provide quick and appropriate real-time remote diagnosis application in classifying PCG of CHD from non-CHD subjects.Clinical Relevance- The suggested methodology can be utilized to analyze PCG signals more quickly and affordably for rural doctors as a first screening tool before sending the cases to experts.

Construction and validation of a nomogram based on the log odds of positive lymph nodes to predict cancer-specific survival in patients with small cell lung cancer after surgery

BackgroundThe lymph node ratio (LNR) is useful for predicting survival in patients with small cell lung cancer (SCLC). The present study compared the effectiveness of the N stage, number of positive LNs (NPLNs), LNR, and log odds of positive LNs (LODDS) to predict cancer-specific survival (CSS) in patients with SCLC. Materials and methods674 patients were screened using the Surveillance Epidemiology and End Results database. The Kaplan-Meier survival and receiver operating characteristic (ROC) curves were performed to address optimal estimation of the N stage, NPLNs, LNR, and LODDS to predict CSS. The optimal LN status group was incorporated into a nomogram to estimate CSS in SCLC patients. The ROC curve, decision curve analysis, and calibration plots were utilized to test the discriminatory ability and accuracy of this nomogram. ResultsThe LODDS model showed the highest accuracy compared to the N stage, NPLNs, and LNR in predicting CSS for SCLC patients. LODDS, age, sex, tumor size, and radiotherapy status were included in the nomogram. The results of calibration plots provided evidences of nice consistency. The ROC and DCA plots suggested a better discriminatory ability and clinical applicability of this nomogram than the 8th TNM and SEER staging systems. ConclusionsLODDS demonstrated a better predictive power than other LN schemes in SCLC patients after surgery. A novel LODDS-incorporating nomogram was built to predict CSS in SCLC patients after surgery, proving to be more precise than the 8th TNM and SEER staging.

Spatial and Temporal Analysis of Climatic Precursors before Major Earthquakes in Iran (2011–2021)

The present study provides a systematic assessment of the relationships between climatic variables and major earthquakes (M &gt; 6) in Iran (2011–2021). These variables include total cloud cover (tcc), low cloud cover (lcc), total precipitation (tp), surface latent heat flux (slhf), and total column rainwater (tcrw). Based on a wider set of variables provided by a multidimensional global dataset (ERA5), the combination of a cross-correlation function (CCF) and receiver operating characteristic (ROC) was used to develop the spatial and temporal analytic relations. Covering maximal values from 0.42 to 0.92, the CCF plots revealed that an increase in climatic parameters could provide valuable information about impending earthquake activity within 8 to 20 days. The mean values of tcc, lcc, tp, slhf, and tcrw were found to increase by 95%, 60%, 80.0 mm, 105 W/m2, and 95 kg−3/m2, respectively. In fact, with the mean AUC (area under the curve) indices ranging from 0.677 (tcc) to 0.810 (tcrw) prior to major earthquakes, the ROC plots allowed for discrimination between seismic and climatic variables ranging from “acceptable” to “excellent”. The changes in the climatic variables under study were due to anomalous air ionization and water condensation in the atmosphere, which can be regarded as short-term precursors to major earthquakes.

An experimental study to diagnose coronary artery disease using five treadmill scores

Objectives: Exercise is the first step of a cardiovascular stress test, including electrocardiography (ECG) and blood pressure monitoring. Typically, a treadmill or bicycle is used to carry out the exercise for this stress test. Computer-derived criteria and a scoring system have been proposed and tested in hopes of improving the diagnostic accuracy of the exercise ECG for diagnosing coronary artery disease (CAD). Many treadmill scores have not been compared with one another in the same population, so questions remain concerning their diagnostic accuracy. The study aimed to correlate the diagnostic accuracy of ST segment response with various treadmill scores. Material and Methods: A total of 100 consecutive patients with suspected CAD referred for exercise testing at ACS Medical College and Hospital, Chennai, Tamil Nadu, between December 2021 and June 2022 were included in the study. The Institutional Review Board approved the study protocol at our institution, and all patients provided written informed consent. A treadmill test was conducted using commercially available equipment. All patients were subjected to symptom-limited exercise testing following the standard Bruce and modified Bruce protocols. A series of five scores were calculated using data collected from the patients to calculate the probability of CAD occurrence. Results: The study included 100 patients with an average age of 48.4 ± 1 years. Obesity (41%), hypertension (80%), diabetes (86%), current smokers (62%), family history of CAD (46%), and hypercholesterolemia (46%) were all observed in the study. The receiver operator characteristic (ROC) plot’s area under the curve (AUC) (standard error) for the Detrano score was 0.46 ± 0.59. This was significantly lower than the AUC of each treadmill score. The AUC of the ROC plots of the Duke treadmill score (DTS), veterans affairs (VA), and consensus scores were 0.47 ± 0.58, 0.63 ± 0.56, and 0.61 ± 0.57, respectively. Conclusion: The Morise score remains helpful in our study population for detecting CAD and determining risk strata. DTS and Detrano had comparable accuracy when calculated from intermediate and high probability scores and ST responses (44% and 43%, respectively). VA and consensus had lower accuracy than others (37% and 29%, respectively).

1186-P: Can an Early GCT Value and A1C Eliminate the Need for Repeat GCT Testing during Second-Trimester Pregnancy?

Background: While A1c is a widely used marker for IDDM and NIDDM, its clinical use in gestational diabetes mellitus (GDM) is not well elucidated. We aimed to evaluate if an A1c and an early glucose challenge test (GCT) can be used to identify patients who may forgo repeat 2nd trimester GCT testing. Methods: This was a retrospective cohort study of patients eligible for early GDM screening who had a negative screening at ≤20 weeks gestation and delivered at a single center between 2016-2020. The primary outcome was GDM in the 2nd trimester by Carpenter &amp; Coustan criteria. Receiver operating characteristic (ROC) curve analyses were performed to compare the ability of early GCT value, and in combination with A1c to predict the primary outcome. Results: Of the 2,025 patients eligible for early GDM screening, 722 met inclusion criteria. Of these, 583 had a 1-hour GCT, 178 had an A1c, and 38 had both. The prediction of GDM diagnosis in the 2nd trimester using an early GCT value, A1c, and both on a ROC plot produced an area under the curve (AUC) of 0.753, 0.659, and 0.672, respectively. The highest A1c where no patients were diagnosed with GDM in 2nd trimester was 4.5%, but of the 43 patients with A1c ≤5.1%, 1 (2.3%) was diagnosed with GDM in the 2nd trimester. Conclusion: Early GCT value alone was a better predictor of GDM in 2nd trimester than a combination of GCT with A1c. A1c alone was a poor predictor. Disclosure J.A.Beacham: None. K.C.Bodycot: None. R.Huang: None. H.B.Al-kouatly: None. R.Mclaren: None.

The Association of IFN-γ, TNF-α, and Interleukins in Bronchoalveolar Lavage Fluid with Lung Cancer: A Prospective Analysis.

Lung cancer is a leading cause of cancer-related mortality worldwide. Identifying novel diagnostic and prognostic biomarkers is essential for improving patient outcomes. This study aimed to investigate the predictive role of cytokines from bronchoalveolar lavage fluid (BALF) in lung cancer diagnosis and prognosis. A prospective study was conducted on 33 patients with suspected lung cancer, divided into inflammatory and non-inflammatory BALF groups. Inflammatory markers in BALF were assessed, and their association with lung cancer risk was analyzed using receiver operating characteristic (ROC) plot analysis, sensitivity and specificity percentages, and regression analysis. Statistically significant differences were observed between the inflammatory and non-inflammatory groups for several inflammatory markers, including IFN-gamma, IL-1b, IL-2, IL-6, IL-10, and IL-12p70. In the follow-up analysis, significant differences persisted for IFN-gamma, IL-1b, IL-2, IL-4, and IL-6. ROC plot analysis revealed that IL-12p70 had the highest area under the curve (AUC) value (0.702), followed by IL-2 (0.682), IL-6 (0.620), IL-4 (0.611), TNF-alpha (0.609), IL-10 (0.604), IL-1b (0.635), and IFN-gamma (0.521). IL-6 showed the highest sensitivity (73%), and IL-1b had the highest specificity (69%). Regression analysis demonstrated that IL-6 (cut-off = 25 pg/mL) and IL-12p70 (cut-off = 30 pg/mL) had the highest odds ratios for lung cancer risk, at 5.09 (95% CI: 2.38-9.24, p < 0.001) and 4.31 (95% CI: 1.85-8.16, p < 0.001), respectively. Cytokines from BALF, particularly IL-6 and IL-12p70, show potential as diagnostic and prognostic biomarkers for lung cancer. Further studies with larger cohorts are warranted to confirm these findings and elucidate the clinical implications of these markers in lung cancer management.

Applications of Natural Language Processing to Geoscience Text Data and Prospectivity Modeling

Geological maps are powerful models for visualizing the complex distribution of rock types through space and time. However, the descriptive information that forms the basis for a preferred map interpretation is typically stored in geological map databases as unstructured text data that are difficult to use in practice. Herein we apply natural language processing (NLP) to geoscientific text data from Canada, the U.S., and Australia to address that knowledge gap. First, rock descriptions, geological ages, lithostratigraphic and lithodemic information, and other long-form text data are translated to numerical vectors, i.e., a word embedding, using a geoscience language model. Network analysis of word associations, nearest neighbors, and principal component analysis are then used to extract meaningful semantic relationships between rock types. We further demonstrate using simple Naive Bayes classifiers and the area under receiver operating characteristics plots (AUC) how word vectors can be used to: (1) predict the locations of “pegmatitic” (AUC = 0.962) and “alkalic” (AUC = 0.938) rocks; (2) predict mineral potential for Mississippi-Valley-type (AUC = 0.868) and clastic-dominated (AUC = 0.809) Zn-Pb deposits; and (3) search geoscientific text data for analogues of the giant Mount Isa clastic-dominated Zn-Pb deposit using the cosine similarities between word vectors. This form of semantic search is a promising NLP approach for assessing mineral potential with limited training data. Overall, the results highlight how geoscience language models and NLP can be used to extract new knowledge from unstructured text data and reduce the mineral exploration search space for critical raw materials.

Modernising Receiver Operating Characteristic (ROC) Curves

The justification for making a measurement can be sought in asking what decisions are based on measurement, such as in assessing the compliance of a quality characteristic of an entity in relation to a specification limit, SL. The relative performance of testing devices and classification algorithms used in assessing compliance is often evaluated using the venerable and ever popular receiver operating characteristic (ROC). However, the ROC tool has potentially all the limitations of classic test theory (CTT) such as the non-linearity, effects of ordinality and confounding task difficulty and instrument ability. These limitations, inherent and often unacknowledged when using the ROC tool, are tackled here for the first time with a modernised approach combining measurement system analysis (MSA) and item response theory (IRT), using data from pregnancy testing as an example. The new method of assessing device ability from separate Rasch IRT regressions for each axis of ROC curves is found to perform significantly better, with correlation coefficients with traditional area-under-curve metrics of at least 0.92 which exceeds that of linearised ROC plots, such as Linacre’s, and is recommended to replace other approaches for device assessment. The resulting improved measurement quality of each ROC curve achieved with this original approach should enable more reliable decision-making in conformity assessment in many scenarios, including machine learning, where its use as a metric for assessing classification algorithms has become almost indispensable.

Benign vs malignant vertebral compression fractures with MRI: a comparison between automatic deep learning network and radiologist's assessment.

To test the diagnostic performance of a deep-learning Two-Stream Compare and Contrast Network (TSCCN) model for differentiating benign and malignant vertebral compression fractures (VCFs) based on MRI. We tested a deep-learning system in 123 benign and 86 malignant VCFs. The median sagittal T1-weighted images (T1WI), T2-weighted images with fat suppression (T2WI-FS), and a combination of both (thereafter, T1WI/T2WI-FS) were used to validate TSCCN. The receiver operator characteristic (ROC) curve was analyzed to evaluate the performance of TSCCN. The accuracy, sensitivity, and specificity of TSCCN in differentiating benign and malignant VCFs were calculated and compared with radiologists' assessments. Intraclass correlation coefficients (ICCs) were tested to find intra- and inter-observer agreement of radiologists in differentiating malignant from benign VCFs. The AUC of the ROC plots of TSCCN according to T1WI, T2WI-FS, and T1WI/T2WI-FS images were 99.2%, 91.7%, and 98.2%, respectively. The accuracy of T1W, T2WI-FS, and T1W/T2WI-FS based on TSCCN was 95.2%, 90.4%, and 96.2%, respectively, greater than that achieved by radiologists. Further, the specificity of T1W, T2WI-FS, and T1W/T2WI-FS based on TSCCN was higher at 98.4%, 94.3%, and 99.2% than that achieved by radiologists. The intra- and inter-observer agreements of radiologists were 0.79-0.85 and 0.79-0.80 for T1WI, 0.65-0.72 and 0.70-0.74 for T2WI-FS, and 0.83-0.88 and 0.83-0.84 for T1WI/T2WI-FS. The TSCCN model showed better diagnostic performance than radiologists for automatically identifying benign or malignant VCFs, and is a potentially helpful tool for future clinical application. TSCCN-assisted MRI has shown superior performance in distinguishing benign and malignant vertebral compression fractures compared to radiologists. This technology has the value to enhance diagnostic accuracy, sensitivity, and specificity. Further integration into clinical practice is required to optimize patient management. • The Two-Stream Compare and Contrast Network (TSCCN) model showed better diagnostic performance than radiologists for identifying benign vs malignant vertebral compression fractures. • The processing of TSCCN is fast and stable, better than the subjective evaluation by radiologists in diagnosing vertebral compression fractures. • The TSCCN model provides options for developing a fully automated, streamlined artificial intelligence diagnostic tool.