Area Under ROC Curve Research Articles

Can resting lung function predict the response of a person living with motor neuron disease to a hypoxic challenge test?

People living with MND (PlwMND) are at risk of altitude-related hypoxia during flight. The Hypoxic Challenge Test (HCT) determines whether in-flight oxygen is required but can be expensive and inaccessible. To assist with travel recommendations, we investigated the relationship between altitude simulation-induced hypoxemia and baseline lung function. Retrospective audit of clinical database of PlwMND who had HCT and lung function. Pearson's correlation assessed relationships between oxygen saturation at altitude (AltSpO2) and lung function. Univariate logistic regression analysis and receiver operator characteristic (ROC) curves determined associations between lung function and HCT pass or fail. Between 2004-2023, 50 PlwMND were identified (median (IQR) diagnosis to HCT = 11.6 (16.9) months, mean ± SD forced vital capacity (FVC) = 2.4 ± 0.9 liters). Ten patients dropped below 85% SpO2 during testing (HCT fail). Baseline SpO2 was associated with AltSpO2 (r = 0.64) and predicted HCT pass or fail (OR 2.0 [95% CI 1.2-3.4], area under ROC curve (AUC) =0.8 [0.6-1.0]), as did FVC (AUC = 0.8 [0.6-0.9]). PlwMND with a FVC > 2.7L or a resting SpO2 > 97% are likely to pass HCT, whereas all those with FVC < 1L and SpO2 < 92% failed. PlwMND with FVC >2.7L or SpO2 >97% are unlikely to require oxygen or ventilatory supports for airline travel. A FVC below 2.7L will require a HCT to confidently determine HCT outcome, with testing still required for FVC <1L or baseline SpO2 <92%, to provide evidence to the airlines for in-flight respiratory support.

Promising biomarker panel to monitor therapeutic efficacy of neoadjuvant chemotherapy in pancreatic cancer patients.

Neoadjuvant chemotherapy (NAC) can provide improved survival outcomes in pancreatic ductal adenocarcinoma (PDAC) patients who respond to treatment, but currently available biomarkers cannot reliably predict NAC response. This study aimed to determine the potential of a previously identified diagnostic and prognostic biomarker panel (i.e. Ca-125, S100A2, S100A4, Mesothelin and Ca19-9) for the monitoring of NAC-response in PDAC patients. This single-centre, retrospective study, utilised serum from NAC treated PDAC patients to determine the levels of biomarkers by Enzyme-Linked Immunosorbent Assay (ELISA). The percentage of the tumour bed occupied by viable carcinoma (PVC) was used to divide patients into good (PVC < 50%) and poor (PVC ≥ 50%) NAC-responders. Statistical analysis was performed to measure the ability of individual biomarkers and a biomarker panel in NAC treatment response and patient survival. Serum specimens from a total of 108 PDAC patients were assessed. Ca-125, Ca19-9 and S100A2 showed a significant positive correlation with PVC. Ca-125 demonstrated a superior ability to monitor NAC treatment response (Area under receiver operating curve (AUC): .6954) compared to the more widely used clinical biomarker, Ca19-9 (AUC: .6291). A panel of Ca-125 and Ca19-9 showed good ability to monitor NAC response in PDAC patients (AUC: .7349). Patients with high levels of both Ca-125 and Ca19-9 were shown to have the poorest overall survival (median overall survival: 17 vs. 30 months). A serum biomarker panel of Ca-125 and Ca19-9 could be used for effective clinical management of PDAC patients undergoing NAC treatment.

Non-invasive lipid panel of MASLD fibrosis transition underscores the role of lipoprotein sulfatides in hepatic immunomodulation.

There exists a pressing need for a non-invasive panel that differentiates mild fibrosis from non-fibrosis in metabolic dysfunction-associated steatotic liver disease (MASLD). In this work, we applied quantitative lipidomics and sterolomics on sera from the PERSONS cohort with biopsy-based histological assessment of liver pathology. We trained a lasso regression model using quantitative omics data and clinical variables, deriving a combinatorial panel of lipids and clinical indices that differentiates mild fibrosis (>F1, n= 324) from non-fibrosis (F0, n= 195), with an area under receiver operating characteristic curve (AUROC) at 0.775 (95% confidence interval [CI]: 0.735-0.816). Circulating sulfatides (SLs) emerged as central lipids distinctly associated with fibrosis pathogenesis in MASLD. Lipidomics analysis of lipoprotein fractions revealed a redistribution of circulating SLs from high-density lipoproteins (HDLs) onto low-density lipoproteins (LDLs) in MASLD fibrosis. We further verified that patient LDLs with reduced SL content triggered a smaller activation of type II natural killer T lymphocytes, compared with control LDLs. Our results suggest that hepatic crosstalk with systemic immunity mediated by lipoprotein metabolism underlies fibrosis progression at early-stage MASLD.

How do we safely preserve ovaries in patients with cervical adenocarcinoma: risk factors and predictive models

ObjectiveTo study and predict the risk of ovarian metastasis (OM) in patients with cervical adenocarcinoma (ADC).MethodsPatients with ADC who received surgical treatment from January 2015 to December 2022 in the Obstetrics and Gynecology Hospital of Fudan University were included in the study. Patients were further divided into OP (ovaries were preserved in surgery) and BSO (bilateral salpingo-oophorectomy) groups. For the patients who accepted BSO, 60% of the patients were randomly grouped into a training cohort, and predictive prognostic models were constructed with 10-fold cross-validation through random forest, LASSO, stepwise, and optimum subset analysis. The model with the highest area under receiver operator curve (AUC) was screened out in the testing cohort. The nomogram and its calibration curve presented the possibility of OM in future patients. The prognoses between the different populations were compared using Kaplan–Meier analysis. All data processing was carried out in R 4.2.0 software.ResultsA total of 934 patients were enrolled in our cohort; 266 patients had their ovaries preserved and 668 patients had BSO according to the previous criteria reported The clinical safety with these criteria was secured, while the 5-year overall survival had no significant difference between the BSO and OP groups (p = 0.069), which suggested that the current criteria could be extended and are more precise. Four predictive models for ovarian metastasis by machine learning were constructed in our study, and the random forest model that obtained the highest AUC in both training and testing sets (0.971 for training and 0.962 for testing set) was taken as the best model. The optimal cut-off point of the ROC curve (specificity 99.5% and 90% sensitivity) was utilized to stratify the patients into high- and low-risk OM. Further comparing the survival curves of the high and low-OM risk groups, it was found that both DFS and OS were significantly prolonged in the low-risk group (p &amp;lt; 0.01). On the basis of this random forest model, a nomogram was used to calculate the OM risk, and the results were validated with calibration. The predictive model was further applied to the whole cohort (934 patients), and we identified the OM low-risk population (n = 822) and the patients with high risk who should be recommended for BSO (n = 112). No significant difference was found in the 5-year survival between the low-risk group with our model and the patients who already had ovaries preserved according to the previous criteria (n = 266).ConclusionThe predictive model constructed in our study could identify the low-risk population of OM in patients with ADC, which remarkably extended the number with the previous criteria, for whom we could potentially preserve ovaries to help improve their life quality.

Machine learning for prediction of incident heart failure and heart failure hospitalisation: multi-cohort development, validation and association with clinical phenotypes

Abstract Background Delayed heart failure (HF) diagnosis is associated with adverse outcomes and costs of hospitalisation.(1,2) Whilst previous tools have been developed to predict incident HF,(3) risk of HF incidence may correlate poorly with risk of HF hospitalisation, which may be more important in targeting diagnostics. Purpose To develop and validate a novel decision support tool for prediction of incident HF and HF hospitalisation, and investigate association of score with HF with reduced ejection fraction (HFrEF) and preserved ejection fraction (HFpEF). Methods We developed the models (FIND-HF) in UK primary care health record data from individuals aged ≥40 years without known HF (Jan 2, 1998 to Feb 28, 2022), randomly divided into training (80%) and testing (20%) datasets. We evaluated logistic regression and supervised machine learning models for prediction of 1- and 5- year HF and HF hospitalisation risk in the testing dataset with internal bootstrap validation. Models were externally validated for HF hospitalisation in data from a Taiwanese university hospital. Association of FIND-HF score with HF phenotypes of HFpEF and HFrEF was assessed in two prospective cohorts, the MATCH registry of patients diagnosed with HFrEF and the FIND-AF cohort of individuals at higher predicted risk of atrial fibrillation who underwent echocardiography. Results Of 565 284 UK individuals (mean age 68.9 (SD 12.2) years, 49.2% women), incidence of HF and HF hospitalisation was 1.1% and 0.5% at 1-year, and 6.6% and 3.4% at 5-years, respectively. Prediction performance was superior for machine learning algorithms compared with logistic regression, with best performance for XGBoost (area under receiver operating characteristic curve (AUROC) at 1 and 5-years: 0.755 and 0.723 for incident HF, and 0.738 and 0.711 for incident HF hospitalization). On external validation in 106,026 individuals in Taiwan (mean age 64.7 (SD 10.6) years, 52.8 % women), HF hospitalisation incidence was 0.71%, and 2.46% at 1- and 5- years, respectively, and discrimination performance was excellent (AUROC at 1- and 5-years: 0.834 and 0.843). When the FIND-HF score was applied to individuals in the MATCH cohort (n = 133, mean age 69.0 (SD 11.7) years, 38.3% women) and FIND-AF cohort (n = 82, mean age 71.6 (SD 7.5) years, 50.0% women), the distribution of score to clinical phenotypes demonstrated a binomial distribution (Figure 1), and likelihood of HFpEF diagnosis correlated with the numeric score (Figure 2). Conclusions The machine learning FIND-HF decision support tool can accurately identify individuals at risk of incident HF and HF hospitalisation, to enable new approaches for early detection and prevention. The variation of score amongst high risk individuals with different HF clinical phenotypes emphasizes the distinct pathophysiological pathways but shared adverse outcome profile across the HF syndrome.Figure 1Figure 2

Cardiac magnetic resonance in MINOCA patients: a non-tertiary hospital experience

Abstract Background Cardiac magnetic resonance (CMR) is a key tool in the diagnostic process of myocardial infarction with non-obstructive coronary arteries (MINOCA), impacting patient management and prognosis. Its use in non-tertiary hospitals is often constrained mainly by accessibility. Thus, the aim of this study was thoroughly evaluate CMR application in this context to optimize the process. Methods 136 MINOCA patients who were admitted to our hospital between 2016 and January 2024 were analyzed. Each MINOCA patient was reviewed by experts to align with the diagnosis criteria outlined in the 2020 ESC Guidelines and the American Heart Association position paper, which recommended CMR for MINOCA patient’s evaluation. To assess the factors associated with normal-CMR, a logistic regression analysis was performed. In a first step, some factors were detected, showing a trend in the association. In a second step, a backward elimination procedure was used, excluding factors with p&amp;gt;0.1. Results CMR was performed in 65 MINOCA patients (48% of the cohort), with a mean age of 59±15 years old, 35 of them were males (54% of total patients). During the follow-up, there were 4 deaths (6%) and 4 readmissions (6%). Out of the total 65 patients, the CMR diagnosis was myocardial infarction (MI) in 14 (21.5%) and myocarditis in 16 (24.6%), and was normal in 35 (53.9%). Therefore, CMR provided diagnostic information in 46% of the cases. It was noted that a normal CMR was associated with female gender, (OR 2.7, 95% CI 0.96-7.34, p=0.057), hypertension (OR 2.25, 95% CI 0.81-6.22, p=0.11), absence of segmental abnormalities in echocardiography (OR 0.14, 95% CI 0.035-0566, p=0.06) and lower high-sensitivity Troponin T (OR 0.79, 95% CI: 0.66-0.95, p=0.001). In the second step, only segmental abnormalities (OR 0.2, 95%CI 0.04-1.09, p=0.063) and troponin (OR 0.31, 95%CI 0.10-0.99, p=0.048) remained included. Additionally, the Youden Index was determined to obtain the threshold for troponin, resulting a value of 606 nG/mL, area under ROC curve (AUC) = 0.784 (figure). Conclusions - CMR was performed in 48% of the MINOCA patients, reflecting the limitation for its use in routine clinical practice. Furthermore, it was useful in 46% of those patients. - The main factors associated with normal CMR were absence of segmental echocardiography abnormalities and reduced troponin levels. - A troponin cutoff point of 606 ng/mL identified patients with a normal CMR (AUC=0.784).Figure 1.CMR in MINOCA patientsFigure 2.Area under ROC curve = 0.784

Association of type-D personality and left-ventricular remodelling in patients treated with primary percutaneous intervention after ST-segment elevation myocardial infarction

BackgroundType-D personality is an established predisposing factor for various diseases. Type-D traits have been shown to pose a 26% increased risk of coronary artery disease after controlling for other confounding factors. Significant associations have been reported between type-D personality traits and dyslipidaemia, impaired endothelial function, coronary heart disease (CAD), acute myocardial infarction, and other adverse cardiovascular events.ObjectiveTo assess the association between type-D personality and left-ventricular adverse remodelling in patients treated with percutaneous coronary intervention following index ST-segment elevation myocardial infarction.MethodsAll patients hospitalized and treated with percutaneous coronary intervention (PCI) after their index ST-segment elevation myocardial infarction (STEMI) between 1 January 2022 to 31 December 2023 were prospectively enrolled. Type-D personality traits in the study population were determined at baseline using type-D Scale-14 (DS14) instrument, whereas any positive change in left ventricular end diastolic volume (LVEDV) ≥ 20% at follow up period of 12-months from baseline was defined as left-ventricular adverse remodelling (LVAR). Univariate and multivariate analysis was done to establish the independent predictors of LVAR. The area under receiver-operating characteristic curve (AUROC) was employed to assess the sensitivity and specificity of the identified independent predictors.ResultsA total of 124 patients were enrolled in the study. The mean age of the study population was 67 ± 10 years and the overall incidence of LVAR was found to be 25%. Multivariate regression analysis revealed that type-D personality is a significant independent predictor of LVAR \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\:(OR:2.51;95\\%CI:1.16-5.77;p=0.03)$$\\end{document} apart from the already established independent predictors Killip Class\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\:\\:(OR:7.30;95\\%CI:3.03-17.55;p<0.0001)$$\\end{document}, baseline Global Longitudinal strain (GLS)\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\:\\:(OR:2.69;95\\%CI:1.19-6.61;p=0.01)$$\\end{document}, and 3-vessel CAD\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\:\\:(OR:2.27;95CI:1.10-5.33;p=0.04)$$\\end{document}. In ROC curve analysis type-D personality as an independent predictor of LVAR achieved a sensitivity of 41.4% and a specificity of 87.1%, p < 0.02.ConclusionType-D personality trait is a significant independent predictor of LVAR in patients treated with PCI after their index-STEMI.

A Strategy for Predicting the Performance of Supervised and Unsupervised Tabular Data Classifiers

AbstractMachine Learning algorithms that perform classification are increasingly been adopted in Information and Communication Technology (ICT) systems and infrastructures due to their capability to profile their expected behavior and detect anomalies due to ongoing errors or intrusions. Deploying a classifier for a given system requires conducting comparison and sensitivity analyses that are time-consuming, require domain expertise, and may even not achieve satisfactory classification performance, resulting in a waste of money and time for practitioners and stakeholders. This paper predicts the expected performance of classifiers without needing to select, craft, exercise, or compare them, requiring minimal expertise and machinery. Should classification performance be predicted worse than expectations, the users could focus on improving data quality and monitoring systems instead of wasting time in exercising classifiers, saving key time and money. The prediction strategy uses scores of feature rankers, which are processed by regressors to predict metrics such as Matthews Correlation Coefficient (MCC) and Area Under ROC-Curve (AUC) for quantifying classification performance. We validate our prediction strategy through a massive experimental analysis using up to 12 feature rankers that process features from 23 public datasets, creating additional variants in the process and exercising supervised and unsupervised classifiers. Our findings show that it is possible to predict the value of performance metrics for supervised or unsupervised classifiers with a mean average error (MAE) of residuals lower than 0.1 for many classification tasks. The predictors are publicly available in a Python library whose usage is straightforward and does not require domain-specific skill or expertise.

Machine Learning For Risk Prediction After Heart Failure Emergency Department Visit or Hospital Admission Using Administrative Health Data.

Patients visiting the emergency department (ED) or hospitalized for heart failure (HF) are at increased risk for subsequent adverse outcomes, however effective risk stratification remains challenging. We utilized a machine-learning (ML)-based approach to identify HF patients at risk of adverse outcomes after an ED visit or hospitalization using a large regional administrative healthcare data system. Patients visiting the ED or hospitalized with HF between 2002-2016 in Alberta, Canada were included. Outcomes of interest were 30-day and 1-year HF-related ED visits, HF hospital readmission or all-cause mortality. We applied a feature extraction method using deep feature synthesis from multiple sources of health data and compared performance of a gradient boosting algorithm (CatBoost) with logistic regression modelling. The area under receiver operating characteristic curve (AUC-ROC) was used to assess model performance. We included 50,630 patients with 93,552 HF ED visits/hospitalizations. At 30-day follow-up in the holdout validation cohort, the AUC-ROC for the combined endpoint of HF ED visit, HF hospital readmission or death for the Catboost and logistic regression models was 74.16 (73.18-75.11) versus 62.25 (61.25-63.18), respectively. At 1-year follow-up corresponding values were 76.80 (76.1-77.47) versus 69.52 (68.77-70.26), respectively. AUC-ROC values for the endpoint of all-cause death alone at 30-days and 1-year follow-up were 83.21 (81.83-84.41) versus 69.53 (67.98-71.18), and 85.73 (85.14-86.29) versus 69.40 (68.57-70.26), for the CatBoost and logistic regression models, respectively. ML-based modelling with deep feature synthesis provided superior risk stratification for HF patients at 30-days and 1-year follow-up after an ED visit or hospitalization using data from a large administrative regional healthcare system.

A computed tomography‑based radio‑clinical model for the prediction of microvascular invasion in gastric cancer.

The objective of the present study was to build and validate a radio-clinical model integrating radiological features and clinical characteristics based on information available before surgery for prediction of microvascular invasion (MI) in gastric cancer. The retrospective study included a cohort of 534 patients (n=374 for the training set and n=160 for the test set) who were diagnosed with gastric cancer. All patients underwent contrast-enhanced computed tomography within one month before surgery. The focal area was mapped by ITK-SNAP. Radiomics features were extracted from portal venous phase CT images. Principal component analysis was used to reduce dimensionality, maximum relevance minimum redundancy, and least absolute shrinkage and selection operator to screen features most associated with MI. The radiomics signature was subsequently computed based on the coefficient weight assigned to it. The independent risk factors for MI of gastric cancer were determined using univariate analysis and multivariate logistic regression analysis. Univariate logistic regression analysis was used to assess the association between clinical characteristics and MI status. A radio-clinical model was constructed by employing multi-variable logistic regression analysis, incorporating radiomic features with clinical characteristics. Receiver operating characteristic (ROC) curve analysis, decision curve analysis (DCA) and calibration curves were employed for the analysis and evaluation of the model's performance. The radiomics signature model had moderate recognition ability, with an area under ROC curve (AUC) of 0.77 for the training set and 0.73 for the test set. The radio-clinical model, consisting of rad-score and clinical features, could well discriminate the training set and test set (AUC=0.88 and 0.80, respectively). The calibration curves and DCA further validated the favorable fit and clinical applicability of the radio-clinical model. In conclusion, the radio-clinical model combining the radiomics signature and clinical characteristics may be used to individually predict MI in gastric cancer to aid in the development of a clinical treatment strategy.

Platelet-to-Lymphocyte Ratio as Coagulopathy Predictor in COVID-19 Patients at Margono Soekarjo Hospital

Coronavirus Disease 2019 (COVID-19) is a disease possibly attacking various organs and systems of the body including the coagulation system causing coagulopathy. Various laboratory biomarkers have been developed to detect coagulopathy. This research aimed to determine the correlation and ability of Platelet-to-Lymphocyte Ratio (PLR) in predicting coagulopathy when compared to D-dimer. A cross-sectional research was conducted on 1580 data of COVID-19 patients at Margono Soekarjo Hospital. The research data were then analyzed using Spearman’s correlation test to figure out the correlation between PLR and D-dimer. This research was also intended to find the PLR Cut-Off Value (COV), Odds Ratio (OR), Area Under Receiver Operating Characteristic (AUROC), and diagnostic value. Platelet-to-lymphocyte ratio value had a significant correlation with D-dimer levels (r=0.260, p=0.000). The COV of PLR was 176.61 with the OR of 2.7 (2.2-3.3 95%CI), AUROC of 0.638, sensitivity of 62%, and specificity of 62%. PLR can be used as a screening biomarker to predict the occurrence of coagulopathy in COVID-19 patients.

Application of serum SAA, CRP, PCT, WBC and N% in the diagnosis of neonatal septicemia

To explore the application value of SAA (serum amyloid A), CRP (C reactive protein), PCT (procalcitonin), WBC (white blood cell) and N% (neutrophil %) in the diagnosis of neonatal septicemia. This study was a retrospective study. 173 children with clinically diagnosed septicemia and 66 children with definitely diagnosed septicemia admitted to the Department of Neonatology, the Affiliated Changsha Central Hospital, Hengyang Medical School, University of South China,from January 2022 to January 2024 were selected as the case group, and 148 children with neonatal jaundice who were hospitalized during the same period were selected as the control group. Fasting venous blood was collected within 24 hours after the children's admission to detect the levels of serum WBC, N%, SAA, CRP and PCT. One-way analysis of variance and Kruskal-Wallis H test were used to compare the general data and inflammatory index levels of the three groups of children. The correlation analysis between SAA and other inflammatory indicators was conducted using Spearman correlation analysis. The receiver operating characteristic (ROC) curve was used to determine the diagnostic efficacy of different inflammatory indicators for patients with definitely diagnosed septicemia and those with clinically diagnosed septicemia, and for those with clinically diagnosed septicemia and those without infection. The results showed that the levels of WBC [(16.88±5.64)×109/L], N% [70.00 (63.00, 75.00)], PCT [2.22 (1.20, 5.55) mg/L], CRP [3.00 (0.50, 10.30) mg/L], SAA [19.70 (10.82, 49.90) mg/L] in the clinically diagnosed septicemia group and WBC [(16.10±7.48)×109/L], N% [73.50 (61.50, 80.93)], PCT [5.35 (0.69, 20.07) mg/L], CRP [15.52 (4.98, 30.50) mg/L], SAA [43.95 (14.00, 175.98) mg/L] in the definitely diagnosed septicemia group were all higher than those in the control group (11.17±3.38)×109/L, 49.81 (36.93, 62.75), 0.20 (0.07, 0.99) mg/L, 0.54 (0.20, 1.40) mg/L, 5.15 (3.60, 8.68) mg/L, and the differences were all statistically significant (all P<0.05). Spearman correlation analysis showed that the level of SAA was positively correlated with WBC, N%, PCT and CRP (rs=0.453, 0.540, 0.343, 0.550, all P<0.05). ROC curve analysis showed that the area under ROC curve(AUC) of SAA for the definitely diagnosed septicemia group and the clinically diagnosed septicemia group was higher than that of other inflammatory indicators, among them, the AUC of SAA for diagnosing the definitely diagnosed neonatal septicemia group was 0.933 (95%CI: 0.809-1.000, P<0.05), with a sensitivity of 92.90% and a specificity of 99.30%. The AUC of SAA for diagnosing the clinically diagnosed septicemia group was 0.861 (95%CI: 0.818-0.904, P<0.05), with a sensitivity of 83.20% and a specificity of 81.80%. In conclusion, compared with CRP, PCT, WBC and N%, SAA has higher sensitivity and specificity for distinguishing neonatal septicemia (including definitely diagnosed septicemia and clinically diagnosed septicemia), and has certain auxiliary diagnostic value for neonatal septicemia.

Development and validation of a nomogram for predicting pulmonary complications in elderly patients undergoing thoracic surgery

BackgroundPostoperative pulmonary complications (PPCs) remain a prevalent concern among elderly patients undergoing surgery, with a notably higher incidence observed in elderly patients undergoing thoracic surgery. This study aimed to develop a nomogram to predict the risk of PPCs in this population.MethodsA total of 2963 elderly patients who underwent thoracic surgery were enrolled and randomly divided into a training cohort (80%, n = 2369) or a validation cohort (20%, n = 593). Univariate and multivariate logistic regression analyses were conducted to identify risk factors for PPCs, and a nomogram was developed based on the findings from the training cohort. The validation cohort was used to validate the model. The predictive accuracy of the model was evaluated by receiver operating characteristic (ROC) curve, area under ROC (AUC), calibration curve, and decision curve analysis (DCA).ResultsA total of 918 (31.0%) patients reported PPCs. Nine independent risk factors for PPCs were identified: preoperative presence of chronic obstructive pulmonary disease (COPD), elevated leukocyte count, higher partial pressure of arterial carbon dioxide (PaCO2) level, surgical site, thoracotomy, intraoperative hypotension, blood loss > 100 mL, surgery duration > 180 min, and malignant tumor. The AUC value for the training cohort was 0.739 (95% CI: 0.719–0.762), and it was 0.703 for the validation cohort (95% CI: 0.657–0.749). The P-values for the Hosmer-Lemeshow test were 0.633 and 0.144 for the training and validation cohorts, respectively, indicating a notable calibration curve fit. The DCA curve indicated that the nomogram could be applied clinically if the risk threshold was between 12% and 84%, which was found to be between 8% and 82% in the validation cohort.ConclusionThis study highlighted the pressing need for early detection of PPCs in elderly patients undergoing thoracic surgery. The nomogram exhibited promising predictive efficacy for PPCs in elderly patients undergoing thoracic surgery, enabling the identification of high-risk patients and consequently aiding in the implementation of preventive interventions.Graphical

GCNPMDA: Human microbe-disease association prediction by hierarchical graph convolutional network with layer attention

Microorganisms play a crucial role in various physiological processes, including metabolism, immune defense, nutrition absorption, defense against cancer, and protection against pathogen colonization. Changes in microbial communities serve as potential biomarkers for diseases, offering significant insights into disease treatment and diagnosis. However, the association between microorganisms and diseases is still unclear, and more computational methods are needed to predict potential associations. In this paper, we introduce a novel computational model, the Graph Convolutional Network to Predict Microbe-Disease Associations (GCNPMDA), which employs layer attention mechanisms (see Figure 1). GCNPMDA integrates known microbe-disease associations, microbe–microbe similarities, and disease–disease similarities into a heterogeneous network. The model utilizes a Graph Convolutional Network (GCN) to learn embeddings for diseases and microbes. To enhance attribute information, microbe–microbe similarities are computed using Cosine similarity, Jaccard similarity, Gaussian kernel, and functional information, while disease–disease similarities are computed using Cosine similarity, Jaccard similarity, Gaussian kernel, and symptom information. Additionally, attention mechanisms are applied to combine embeddings from multiple graph convolution layers. The model’s predictive effectiveness is evaluated on Human Microbe-Disease Association Database (HMDAD). Leave-one-out cross-validation (LOOCV) was conducted. The Area Under ROC Curve (AUC) of LOOCV is 0.98. The 5-fold cross-validation (5-fold CV) on HMDAD yields average AUC of 0.98 ± 0.009. Furthermore, we carried out a case study of type 2 diabetes (T2D), inflammatory bowel disease (IBD), and rheumatoid arthritis. Based on existing literature evidence, it was confirmed that 6, 7, and 7 of the top-10 inferred microbes have established associations with T2D, IBD, and rheumatoid arthritis, respectively. GCNPMDA demonstrates potential efficacy in identifying disease-related microbes, offering a promising tool to uncover the intricate relationship between microorganisms and their human hosts.

A-080 Morning Cortisol for Rule-in and Rule-out of Adrenal Insufficiency in an Asian Population

Abstract Background Short ACTH stimulation (Synacthen) tests are performed to investigate for adrenal insufficiency. Previous work suggests that morning cortisol concentrations can be used to rule in or out adrenal insufficiency to avoid performance of a Synacthen test. This study aimed to determine the optimal cortisol concentration for rule-in or rule-out adrenal insufficiency based on 30-minute cortisol (COR30) for adequate response using the assay-specific cut-off from The Endocrine Society of Australia/The Australasian Association for Clinical Biochemistry and Laboratory Medicine/The Royal College of Pathologists of Australasia (ESA/AACB/RCPA) Harmonization of Endocrine Dynamic Testing - Adult (HEDTA), in an Asian population. Methods Synacthen tests performed at our institution from 2012-2024, with a baseline cortisol between 6 and 10 am (COR0(6-10)), were extracted from the laboratory database. Plasma cortisol concentrations were measured by Beckman Coulter DxI 800. Synacthen tests were classified as adequate or inadequate based on the Beckman Coulter specific HEDTA cut-off for COR30 of 420 nmol/L. Performance characteristics (sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and area under receiver operator characteristic (AUROC) curve) of COR0(6-10) concentrations to predict adequate or inadequate response were calculated using Microsoft Excel. Results 4609 Synacthen tests were included, 77% of which had adequate Synacthen response. COR0(6-10) &amp;gt;300 nmol/L could exclude inadequate COR30 response with sensitivity 95%, specificity 48%, PPV 35%, NPV 97%; and COR0(6-10) &amp;gt;364 nmol/L had sensitivity 99%, specificity 29%, PPV 29%, NPV 99%. COR0(6-10) &amp;lt;155 nmol/L could rule-in inadequate COR30 response with sensitivity 53%, specificity 95%, PPV 76%, NPV 87%; and COR0(6-10) &amp;lt;97 nmol/L had sensitivity 35%, specificity 99%, PPV 91%, NPV 84%. AUROC of COR0(6-10) for adequate COR30 response was 0.86. If patients with COR0(6-10) &amp;lt;97 nmol/L and &amp;gt;300 nmol/L did not proceed to Synacthen testing, up to 46% of tests could be avoided, while keeping misclassification under 5%. Conclusions Morning cortisol cut-offs were determined for rule-in and rule-out of adrenal insufficiency in an Asian population. Patients with morning cortisol &amp;gt;300 nmol/L or &amp;lt;97 nmol/L may be able to avoid Synacthen testing unless there is a strong suspicion of clinical discordance. Such a testing strategy could significantly decrease the number of Synacthen tests while keeping misclassification to a minimum.

B-065 Comparative Analysis of 14-3-3η Autoantibody Assays on Luminex Platform for Diagnosing Axial Spondyloarthritis

Abstract Background The pathophysiological significance of extracellular 14-3-3η antigen and its autoantibodies (AAbs) is well-documented in autoimmune rheumatologic disorders. Use of 14-3-3η protein blood tests is common in diagnosing and managing rheumatoid arthritis (RA). Prior studies have established the diagnostic and prognostic value of 14-3-3η AAbs in RA and axial spondyloarthritis (AS), particularly in differentiating AS patients from healthy individuals and in relation to disease severity and progression. This study explores the performance of a novel assay platform in the context of AS diagnosis, especially in younger patients who are most responsive to early treatment interventions.This research aims to evaluate the discriminative accuracy of the 14-3-3η AAb assay on the Luminex® platform, a widely recognized clinical tool, in distinguishing AS patients from healthy subjects, and to compare its performance with that of the Meso Scale Discovery® (MSD) research-use only (RUO) platform. Methods Sera from 49 AS patients and 25 healthy controls (HC), previously assessed using the MSD platform, were reanalyzed using a new Luminex 14-3-3η AAb assay. These samples were sourced from the Follow Up Research Cohort in Axial Spondyloarthritis (FORCAST) Prognostic Cohort. Spearman correlations were employed for variable analysis, and the Mann-Whitney U-test was used to compare AAb levels between groups. Receiver Operating Characteristic (ROC) Area Under Curve (AUC) analyses were conducted for both assay platforms, with a specific focus on patients aged 45 and under. Results The AS patient group had a mean age (SD) of 44 (13) years, with 67% male representation. The median (IQR) disease duration was 19 (10-24) years, and median (IQR) C-reactive protein (CRP) levels were 9.6 (4.3 - 24.8) mg/L. Median (IQR) 14-3-3η AAb levels were significantly higher in AS patients at 136 U/μl (48 - 408) compared to HCs at 54 (28 - 161) U/μl, p=0.02. ROC AUC analysis for the MSD RUO platform showed an AUC of 0.67 (95% CI, 0.55 - 0.79), and for the Luminex platform, an AUC of 0.66 (95% CI, 0.53 - 0.80), indicating comparable performance across platforms. In the subgroup of AS patients aged ≤45 years, the Luminex assay demonstrated an AUC of 0.74 (95% CI, 0.61-0.88), p=0.003. Combination of 14-3-3η AAbs and CRP was able to identify 90% of AS patients. No correlation between the two markers was observed (r= -0.14, p=0.38). Conclusions The Luminex-based 14-3-3η AAb assay exhibits equivalent discriminative capability to the MSD RUO assay in differentiating AS patients from healthy individuals. The combined use of 14-3-3η AAbs and CRP, identifying 90% of AS patients, highlights the potential of these biomarkers as complementary diagnostic and prognostic tools in AS, meriting further research.

Automated Identification of HeartFailureWith Reduced Ejection Fraction Using Deep Learning-Based Natural Language Processing.

The lack of automated tools for measuring care quality limits the implementation of a national program to assess guideline-directed care in heart failure with reduced ejection fraction (HFrEF). The authors aimed to automate the identification of patients with HFrEF at hospital discharge, an opportunity to evaluate and improve the quality of care. The authors developed a novel deep-learning language model for identifying patients with HFrEF from discharge summaries of hospitalizations with heart failure at Yale New Haven Hospital during 2015 to 2019. HFrEF was defined by left ventricular ejection fraction<40% on antecedent echocardiography. The authors externally validated the model at Northwestern Medicine, community hospitals of Yale, and the MIMIC-III (Medical Information Mart for Intensive Care III) database. A total of 13,251 notes from 5,392 unique individuals (age 73 ± 14 years, 48% women), including 2,487 patients with HFrEF (46.1%), were used for model development (train/held-out: 70%/30%). The model achieved an area under receiver-operating characteristic curve (AUROC) of 0.97 and area under precision recall curve (AUPRC) of 0.97 in detecting HFrEF on the held-out set. The model had high performance in identifying HFrEF with AUROC=0.94 and AUPRC=0.91 on 19,242 notes from Northwestern Medicine, AUROC=0.95 and AUPRC=0.96 on 139 manually abstracted notes from Yale community hospitals, and AUROC=0.91 and AUPRC=0.92 on 146 manually reviewed notes from MIMIC-III. Model-based predictions of HFrEF corresponded to a net reclassification improvement of 60.2 ± 1.9% compared with diagnosis codes (P< 0.001). The authors developed a language model that identifies HFrEF from clinical notes with high precision and accuracy, representing a key element in automating quality assessment for individuals with HFrEF.

Feasibility of Fall-Risk Detection in Older Adults: Real-World Use of Sensor Data With Machine Learning.

To use machine learning techniques with sensor data to predict fall risk in older adults aging in place. We tested the feasibility of using anomaly detection on a dataset comprising 315 days of continuous unobtrusive sensor data obtained from a single participant to predict fall risk within a 10-day window. Predictions were validated with performance metrics, including accuracy, F1 score, and receiver operating characteristic-area under curve (ROC-AUC), using actual falls documented in the electronic health record. The model resulted with accuracy = 0.96 (95% confidence interval [CI] [0.94, 0.99]), F1 = 0.78 (95% CI [0.73, 0.83]), and ROC-AUC = 0.89 (95% CI [0.85, 0.93]). The application of anomaly detection on sensor data may provide a timely and valid indication of fall risk in older adults within a 10-day window. Further research and validation are warranted to confirm these findings and expand the scope of application in the domain of older adult care and health care support. [Journal of Gerontological Nursing, 50(10), 7-10.].

Gut metagenome-derived image augmentation and deep learning improve prediction accuracy of metabolic disease classification.

In recent years, statistics and machine learning methods have been widely used to analyze the relationship between human gut microbial metagenome and metabolic diseases, which is of great significance for the functional annotation and development of microbial communities. In this study, we proposed a new and scalable framework for image enhancement and deep learning of gut metagenome, which could be used in the classification of human metabolic diseases. Each data sample in three representative human gut metagenome datasets was transformed into image and enhanced, and put into the machine learning models of logistic regression (LR), support vector machine (SVM), Bayesian network (BN) and random forest (RF), and the deep learning models of multilayer perceptron (MLP) and convolutional neural network (CNN). The accuracy performance of the overall evaluation model for disease prediction was verified by accuracy (A), accuracy (P), recall (R), F1 score (F1), area under ROC curve (AUC) and 10 fold cross-validation. The results showed that the overall performance of MLP model was better than that of CNN, LR, SVM, BN, RF and PopPhy-CNN, and the performance of MLP and CNN models was further improved after data enhancement (random rotation and adding salt-and-pepper noise). The accuracy of MLP model in disease prediction was further improved by 4%-11%, F1 by 1%-6% and AUC by 5%-10%. The above results showed that human gut metagenome image enhancement and deep learning could accurately extract microbial characteristics and effectively predict the host disease phenotype. The source code and datasets used in this study can be publicly accessed in https://github.com/HuaXWu/GM_ML_Classification.git.

Detection of COVID-19 from chest X-rays using deep transfer learning

Objectives: As COVID-19 continues to wreak havoc around the world, effective screening methods for safeguarding public health and bringing the pandemic under control are urgently required. Accordingly, this study proposes a robust pipeline for training and evaluating deep learning predictive models for the detection of COVID-19 from chest X-rays. Methods: The pipeline incorporates multiple techniques to combat overfitting and optimize the prediction results, including data augmentation, Bayesian optimization, the selection of appropriate performance metrics, and the use of early stopping during model training. The proposed pipeline is applied to three common deep learning models: ResNet50, NASNet-A-Mobile, and Xception. Results: The experimental results obtained using the COVID-XRay-5K v3 dataset with 2084 training and 3100 test examples show that the three models achieve AU-PRC (area under precision-recall curve) scores of 0.977, 0.963 and 0.900, respectively; and AU-ROC (area under receiver operating characteristic) scores of 0.994, 0.996 and 0.981. Moreover, at a 98.00% recall (sensitivity), the models achieve high specificities of 97.53%, 97.60% and 86.00%, respectively. Conclusion: Overall, the results suggest that all three models provide a promising approach for the rapid and reliable detection of COVID-19.