Receiver Operating Characteristic AUC Research Articles

The effect of cognitive function and central nervous system depressant use on mortality-A prospective observational study of previously hospitalised older patients.

Older patients are often users of prolonged Central Nervous System Depressants (CNSD) (Z-hypnotics, benzodiazepines and opioids), which may be associated with reduced cognition. The long-term effects of CNSD use and reduced cognitive function in older patients are unclear. The aim of this study was to examine whether cognitive function and CNSD use at baseline hospitalisation were associated with all-cause mortality two years after discharge. We conducted a prospective observational study, including baseline data (2017-2018) from previously hospitalised older patients (65-90 years), assessing all-cause mortality two years after discharge. We used logistic regression to assess the primary outcome, all-cause mortality two years after baseline hospitalisation. The primary predictors were cognitive function measured by The Mini Mental State Examination (MMSE) and prolonged CNSD use (continuous use ≥ 4 weeks). Adjustment variables: age, gender, education, the Hospital Anxiety and Depression Scale (HADS) and the Cumulative Illness Rating Scale for Geriatrics (CIRS-G), using receiver operating characteristics (ROC) to compare the predictive power of the models. In a sub-analysis we used, the Neurobehavioural Cognitive State Examination (Cognistat) and the Clock Drawing Test. Two years after discharge, out of 246 baseline patients, 43 were deceased at follow-up, among these 27 (63%) were CNSD users, and 16 (36%) were non-users at baseline, (p = 0.002). In the multivariable models cognitive function (MMSE score) was a predictor of mortality (OR 0.81 (95% CI 0.69; 0.96), p = 0.014). CNSD use was associated with mortality (OR 2.71 (95% CI 1.06; 6.95), p = 0.038), with ROC AUC: 0.74-0.77 for these models. Results using Cognistat supported the findings. The Clock Drawing Test was not significant predictor of mortality. Two years after discharge from the hospital, older patients with reduced cognitive function and CNSD use during hospital stay had higher mortality. This underlines that inappropriate (prolonged and concurrent) use of CNSDs should be avoided by older patients, particularly in patients with reduced cognitive function. NCT03162081, 22 May 2017.

Diagnostic performance of ultrasound hepatorenal index for the diagnosis of hepatic steatosis in children.

Non-alcoholic fatty liver disease (NAFLD) is increasing in prevalence and is the most common cause of pediatric chronic liver disease. Objective US-based measures of hepatic steatosis are an unmet clinical need. To evaluate the diagnostic performance of quantitative measurement of liver echogenicity (hepatorenal index, or HRI) for hepatic steatosis in a pediatric cohort. We identified pediatric patients (≤18years old) who underwent both clinically indicated abdominal US and MRI with liver proton-density fat fraction (PDFF) within the 3-month period during the timeframe of July 2015-April 2020 (n=69). Using ImageJ, we drew small circular regions of interest (ROIs) and large freehand ROIs in the liver and right kidney on single longitudinal and transverse images to measure echogenicity (arbitrary units). We calculated four HRIs (liver-to-kidney ratio) as well as liver histogram features. Five pediatric radiologists independently reported the qualitative presence/absence of hepatic steatosis. We used Pearson correlation (r) to assess associations and receiver operating characteristic (ROC) curve analyses to evaluate diagnostic performance. Multivariable logistic regression was used to further assess relationships. Mean patient age was 11.6 (standard deviation [SD] 4.7, range 0.3-18) years; 27/69 (39.1%) were female. Mean PDFF was 12.5% (SD 13.1%, range 1-48%); 34/69 (49.3%) patients were classified as having hepatic steatosis by MRI (PDFF ≥6%). There were significant, positive correlations between all four US HRI methods and PDFF (r=0.51-0.61); longitudinal freehand ROIs exhibited the strongest correlation (r=0.61; P<0.0001). Longitudinal freehand ROI HRI had moderate diagnostic performance for the binary presence of steatosis (area under the curve [AUC]=0.80, P<0.0001), with an optimal cut-off value >1.75 (sensitivity=70.6%, specificity=77.1%). Radiologists' sensitivity for detecting hepatic steatosis ranged from 79.4% to 97.1%, and specificity ranged from 91.2% to 100%. Significant multivariable predictors of PDFF ≥6% included HRI (P=0.002; odds ratio [OR]=34.2), body mass index (BMI) percentile (P=0.005; OR=1.06), and liver gray-scale echogenicity standard deviation (P=0.02; OR=0.79) (receiver operating characteristic AUC = 0.92). Quantitative US HRI has moderate diagnostic performance for detecting liver fat in children and positively correlates with MRI PDFF. Incorporation of BMI-percentile and gray-scale echogenicity standard deviation improved diagnostic performance.

Analysis of serum lipid parameters predicting lipid metabolic disorders in TSC-AML patients with treatment of mTOR inhibitors.

Patients with tuberous sclerosis complex (TSC) demonstrate disrupted lipid homeostasis before and during treatment with mammalian target of rapamycin (mTOR) inhibitor. However, few previous reports focused on if the serum lipid status at baseline would influence lipid metabolic side-effects of mTOR inhibitors for TSC associated renal angiomyolipomas (TSC-AML). The present study was designed to evaluate the predictive function of serum lipid status at baseline for hyperlipidaemia by mTOR inhibitor treatment in TSC-AML patients. The clinical data of TSC-AML patients who took mTOR inhibitors in Department of Urology of Peking Union Medical College Hospital (PUMCH) from 1January 2014 to 1January 2021, were retrospectively analysed. The record of lipid parameters at baseline and the highest levels of total cholesterol (TC) and triglyceride (TG) after treatment at least ≥3months were collected. The correlation of serum lipid parameters at baseline with incidence of hyperlipidaemia during mTOR inhibitor treatment was analysed. Receiver operating characteristic (ROC) curve analysis was conducted to evaluate the ability of the serum lipid parameters in predicting hyperlipidaemia. 19 patients experienced hyperlipidaemia and 13 patients still had normal TC and TG levels during mTOR inhibitor treatment. The levels of high-density lipoprotein cholesterol (HDL-C) (0.98±0.30mmol/L vs. 1.23±0.31mmol/L, p=0.030), low-density lipoprotein cholesterol (LDL-C) (2.47±0.69mmol/L vs. 1.95±0.53mmol/L, p=0.029) and apolipoprotein B (ApoB) (0.82±0.21g/L vs. 0.65±0.16g/L, p=0.019) are higher in the patients who experienced hyperlipidaemia during mTOR inhibition therapy. TC, TG, LDL-C, ApoB and high-sensitivity C-reactive protein (hsCRP) at baseline had positive correlation with TC after treatment; ApoB at baseline had positive correlation, while HDL-C and free fat acid (FFA) at baseline had negative correlation with TG after treatment. Therefore, ApoB concentration at baseline has statistically significant correlation with both TC (p<0.001) and TG (p=0.012) levels after mTOR inhibitor treatment. ROC curve and AUC revealed that ApoB with a cut-off value of 0.640g/L may be the best parameter for predicting hyperlipidaemia during mTOR inhibitor treatment in TSC-AML patients. The incidence rates of hyperlipidaemia were 27.3% and 76.2% among the patients with ApoB level ≤0.640g/L and >0.640g/L respectively. Some baseline serum lipid parameters could be used for predicting incidence of hyperlipidaemia during mTOR inhibition therapy in TSC-AML patients, and ApoB with 0.640g/L as a cut-off value may be a potentially optimal indicator, which could help for diagnosis and treatment decision-making.

Non-invasive precise staging of liver fibrosis using deep residual network model based on plain CT images.

The aim of this study was to explore the application of five-class deep residual network models based on plain CT images and clinical features for the precise staging of liver fibrosis. This retrospective clinical study included 347 patients who underwent liver CT, with pathological staging of liver fibrosis as the gold standard. We established three ResNet models to stage liver fibrosis. The output diagnosis labels of models were 0, 1, 2, 3 and 4, which correspond to F0, F1, F2, F3, and F4 stages. Confusion matrices were used to evaluate the performances of models to precisely stage liver fibrosis. The performance for diagnosing cirrhosis (F4), advanced fibrosis (≥ F3) and significant fibrosis (≥ F2) of models was evaluated with receiver operating characteristic (ROC) analyses. The kappa coefficients of the five-class ResNet model (based on plain CT images), the five-class ResNet clinical model (based on clinical features), and the five-class mixed ResNet model (based on plain CT images and clinical features) for precise staging liver fibrosis were 0.566, 0.306, and 0.63, respectively. The recall rates and precision rates for F0, F1, F2, and F3 of three models were lower than 60%. The ROC AUC values of the five-class ResNet model, the five-class ResNet clinical model, and the five-class mixed ResNet model for diagnosing cirrhosis, advanced fibrosis, and significant fibrosis were 0.95, 0.88, and 0.82, 0.80, 0.72, and 0.70, 0.95, 0.90, and 0.83, respectively. The five-class ResNet models are of high value in the diagnosis of liver cirrhosis, advanced liver fibrosis, and significant liver fibrosis. However, for the precise staging of liver fibrosis, the models cannot accurately distinguish other liver fibrosis stages except F4. Plain CT images combined with clinical features have the potential to improve the performance of the ResNet models in diagnosing liver fibrosis.

Machine learning-based approaches for identifying human blood cells harboring CRISPR-mediated fetal chromatin domain ablations

Two common hemoglobinopathies, sickle cell disease (SCD) and β-thalassemia, arise from genetic mutations within the β-globin gene. In this work, we identified a 500-bp motif (Fetal Chromatin Domain, FCD) upstream of human ϒ-globin locus and showed that the removal of this motif using CRISPR technology reactivates the expression of ϒ-globin. Next, we present two different cell morphology-based machine learning approaches that can be used identify human blood cells (KU-812) that harbor CRISPR-mediated FCD genetic modifications. Three candidate models from the first approach, which uses multilayer perceptron algorithm (MLP 20-26, MLP26-18, and MLP 30-26) and flow cytometry-derived cellular data, yielded 0.83 precision, 0.80 recall, 0.82 accuracy, and 0.90 area under the ROC (receiver operating characteristic) curve when predicting the edited cells. In comparison, the candidate model from the second approach, which uses deep learning (T2D5) and DIC microscopy-derived imaging data, performed with less accuracy (0.80) and ROC AUC (0.87). We envision that equivalent machine learning-based models can complement currently available genotyping protocols for specific genetic modifications which result in morphological changes in human cells.

Identification of Epithelial Mesenchymal Transition-Related lncRNAs Associated with Prognosis and Tumor Immune Microenvironment of Hepatocellular Carcinoma.

Purpose The long noncoding RNAs (lncRNAs) play the important role in tumor occurrence and progression, and the epithelial to mesenchymal transition (EMT) is the critical process for tumor migration. However, the role of EMT-related lncRNA in hepatocellular carcinoma (HCC) has not been elucidated. Methods In this study, we selected the EMT-related lncRNAs in HCC by using data from The Cancer Genome Atlas database (TCGA). Two prognostic models of the overall survival (OS) and relapse-free survival (RFS) were constructed and validated through Cox regression model, Kaplan-Meier analysis, and the receiver-operating characteristic (ROC) curves. The unsupervised clustering analysis was utilized to investigate the association between EMT-lncRNAs with tumor immune microenvironment. ESTIMATE algorithm and gene set enrichment analysis (GSEA) were used to estimate tumor microenvironment and associated KEGG pathways. Results Two EMT-related lncRNA prognostic models of OS and RFS were constructed. Kaplan-Meier curves showed the dismal prognosis of OS and RFS in the group with high-risk score. The ROC curves and AUC values in two prognostic models indicated the discriminative value in the training set and validation set. Patients with HCC were clustered into two subgroups according the unsupervised clustering analysis. Lnc-CCNY-1 was selected as the key lncRNA. GSVA analysis showed that lnc-CCNY-1 was negatively associated with peroxisome proliferator-activated receptor (PPAR) signaling pathway and positively correlated with CELL cycle pathway. Conclusion Two EMT-related lncRNA prognostic models of OS and RFS were constructed to discriminate patients and predict prognosis of HCC. EMT-related lncRNAs may play a role on prognosis of HCC by influencing the immune microenvironment. Lnc-CCNY-1 was selected as the key EMT-related lncRNA for further exploration.

Prediction of Early Onset Cytokine Release Syndrome and Neurologic Events after Axicabtagene Ciloleucel in Large B-Cell Lymphoma Based on Machine Learning Algorithms

Prediction of Early Onset Cytokine Release Syndrome and Neurologic Events after Axicabtagene Ciloleucel in Large B-Cell Lymphoma Based on Machine Learning Algorithms

Metabolomic identification of α-ketoglutaric acid elevation in pediatric chronic graft-versus-host disease

AbstractChronic graft-versus-host disease (cGVHD) is the most common cause for non-relapse mortality postallogeneic hematopoietic stem cell transplant (HSCT). However, there are no well-defined biomarkers for cGVHD or late acute GVHD (aGVHD). This study is a longitudinal evaluation of metabolomic patterns of cGVHD and late aGVHD in pediatric HSCT recipients. A quantitative analysis of plasma metabolites was performed on 222 evaluable pediatric subjects from the ABLE/PBMTC1202 study. We performed a risk-assignment analysis at day + 100 (D100) on subjects who later developed either cGVHD or late aGVHD after day 114 to non-cGVHD controls. A second analysis at diagnosis used fixed and mixed multiple regression to compare cGVHD at onset to time-matched non-cGVHD controls. A metabolomic biomarker was considered biologically relevant only if it met all 3 selection criteria: (1) P ≤ .05; (2) effect ratio of ≥1.3 or ≤0.75; and (3) receiver operator characteristic AUC ≥0.60. We found a consistent elevation in plasma α-ketoglutaric acid before (D100) and at the onset of cGVHD, not impacted by cGVHD severity, pubertal status, or previous aGVHD. In addition, late aGVHD had a unique metabolomic pattern at D100 compared with cGVHD. Additional metabolomic correlation patterns were seen with the clinical presentation of pulmonary, de novo, and progressive cGVHD. α-ketoglutaric acid emerged as the single most significant metabolite associated with cGVHD, both in the D100 risk-assignment and later diagnostic onset analysis. These distinctive metabolic patterns may lead to improved subclassification of cGVHD. Future validation of these exploratory results is needed. This trial was registered at www.clinicaltrials.gov as #NCT02067832.

Predicting the Disease Outcome in COVID-19 Positive Patients Through Machine Learning: A Retrospective Cohort Study With Brazilian Data.

The first officially registered case of COVID-19 in Brazil was on February 26, 2020. Since then, the situation has worsened with more than 672, 000 confirmed cases and at least 36, 000 reported deaths by June 2020. Accurate diagnosis of patients with COVID-19 is extremely important to offer adequate treatment, and avoid overloading the healthcare system. Characteristics of patients such as age, comorbidities and varied clinical symptoms can help in classifying the level of infection severity, predict the disease outcome and the need for hospitalization. Here, we present a study to predict a poor prognosis in positive COVID-19 patients and possible outcomes using machine learning. The study dataset comprises information of 8, 443 patients concerning closed cases due to cure or death. Our experimental results show the disease outcome can be predicted with a Receiver Operating Characteristic AUC of 0.92, Sensitivity of 0.88 and Specificity of 0.82 for the best prediction model. This is a preliminary retrospective study which can be improved with the inclusion of further data. Conclusion: Machine learning techniques fed with demographic and clinical data along with comorbidities of the patients can assist in the prognostic prediction and physician decision-making, allowing a faster response and contributing to the non-overload of healthcare systems.

Toward Guided Mutagenesis: Gaussian Process Regression Predicts MHC Class II Antigen Mutant Binding.

Antigen-specific immunotherapies (ASI) require successful loading and presentation of antigen peptides into the major histocompatibility complex (MHC) binding cleft. One route of ASI design is to mutate native antigens for either stronger or weaker binding interaction to MHC. Exploring all possible mutations is costly both experimentally and computationally. To reduce experimental and computational expense, here we investigate the minimal amount of prior data required to accurately predict the relative binding affinity of point mutations for peptide-MHC class II (pMHCII) binding. Using data from different residue subsets, we interpolate pMHCII mutant binding affinities by Gaussian process (GP) regression of residue volume and hydrophobicity. We apply GP regression to an experimental data set from the Immune Epitope Database, and theoretical data sets from NetMHCIIpan and Free Energy Perturbation calculations. We find that GP regression can predict binding affinities of nine neutral residues from a six-residue subset with an average R2 coefficient of determination value of 0.62 ± 0.04 (±95% CI), average error of 0.09 ± 0.01 kcal/mol (±95% CI), and with an receiver operating characteristic (ROC) AUC value of 0.92 for binary classification of enhanced or diminished binding affinity. Similarly, metrics increase to an R2 value of 0.69 ± 0.04, average error of 0.07 ± 0.01 kcal/mol, and an ROC AUC value of 0.94 for predicting seven neutral residues from an eight-residue subset. Our work finds that prediction is most accurate for neutral residues at anchor residue sites without register shift. This work holds relevance to predicting pMHCII binding and accelerating ASI design.

Presence of SARS-CoV-2 Antibodies in Spent Peritoneal Dialysate.

Presence of SARS-CoV-2 Antibodies in Spent Peritoneal Dialysate.

Predicting local failure of brain metastases after stereotactic radiosurgery with radiomics on planning MR images and dose maps.

Stereotactic radiosurgery (SRS) has become an important modality in the treatment of brain metastases. The purpose of this study is to investigate the potential of radiomic features from planning magnetic resonance (MR) images and dose maps to predict local failure after SRS for brain metastases. Twenty-eight patients who received Gamma Knife (GK) radiosurgery for brain metastases were retrospectively reviewed in this IRB-approved study. 179 irradiated tumors included 42 that locally failed within one-year follow-up. Using SRS tumor volumes, radiomic features were calculated on T1-weighted contrast-enhanced MR images acquired for treatment planning and planned dose maps. 125 radiomic features regarding tumor shape, dose distribution, MR intensities and textures were extracted for each tumor. Logistic regression with automatic feature selection was built to predict tumor progression from local control after SRS. Feature selection and model evaluation using receiver operating characteristic (ROC) curves were performed in a nested cross validation (CV) scheme. The associations between selected radiomic features and treatment outcomes were statistically assessed by univariate analysis. The logistic model with feature selection achieved ROC AUC of 0.82±0.09 on 5-fold CV, providing 83% sensitivity and 70% specificity for predicting local failure. A total of 10 radiomic features including 1 shape feature, 6 MR images and 3 dose distribution features were selected. These features were significantly associated with treatment outcomes (p<0.05). The model was validated on independent holdout data with an AUC of 0.78. Radiomic features from planning MR images and dose maps provided prognostic information in SRS for brain metastases. A model built on the radiomic features shows promise for early prediction of tumor local failure after treatment, potentially aiding in personalized care for brain metastases.

Construction of a novel mRNA-signature prediction model for prognosis of bladder cancer based on a statistical analysis

BackgroundBladder cancer (BC) is a common malignancy neoplasm diagnosed in advanced stages in most cases. It is crucial to screen ideal biomarkers and construct a more accurate prognostic model than conventional clinical parameters. The aim of this research was to develop and validate an mRNA-based signature for predicting the prognosis of patients with bladder cancer.MethodsThe RNA-seq data was downloaded from the Cancer Genome Atlas (TCGA) and Gene Expression Omnibus (GEO). Differentially expressed genes (DEGs) were screened in three datasets, and prognostic genes were identified from the training set of TCGA dataset. The common genes between DEGs and prognostic genes were narrowed down to six genes via Least Absolute Shrinkage and Selection Operator (LASSO) regression, and stepwise multivariate Cox regression. Then the gene-based risk score was calculated via Cox coefficient. Time-dependent receiver operating characteristic (ROC) and Kaplan-Meier (KM) survival analysis were used to assess the prognostic power of risk score. Multivariate Cox regression analysis was applied to construct a nomogram. Decision curve analysis (DCA), calibration curves, and time-dependent ROC were performed to assess the nomogram. Finally, functional enrichment of candidate genes was conducted to explore the potential biological pathways of candidate genes.ResultsSORBS2, GPC2, SETBP1, FGF11, APOL1, and H1–2 were screened to be correlated with the prognosis of BC patients. A nomogram was constructed based on the risk score, pathological stage, and age. Then, the calibration plots for the 1-, 3-, 5-year OS were predicted well in entire TCGA-BLCA patients. Decision curve analysis (DCA) indicated that the clinical value of the nomogram was higher than the stage model and TNM model in predicting overall survival analysis. The time-dependent ROC curves indicated that the nomogram had higher predictive accuracy than the stage model and risk score model. The AUC of nomogram time-dependent ROC was 0.763, 0.805, and 0.806 for 1-year, 3-year, and 5-year, respectively. Functional enrichment analysis of candidate genes suggested several pathways and mechanisms related to cancer.ConclusionsIn this research, we developed an mRNA-based signature that incorporated clinical prognostic parameters to predict BC patient prognosis well, which may provide a novel prognosis assessment tool for clinical practice and explore several potential novel biomarkers related to the prognosis of patients with BC.

Clinical assessment of a biophysical model for distinguishing tumor progression from radiation necrosis

The efficacy of an imaging-driven mechanistic biophysical model of tumor growth for distinguishing radiation necrosis from tumor progression in patients with enhancing lesions following stereotactic radiosurgery (SRS) for brain metastasis is validated. We retrospectively assessed the model using 73 patients with 78 lesions and histologically confirmed radiation necrosis or tumor progression. Postcontrast T1-weighted MRI images were used to extract parameters for a mechanistic reaction-diffusion logistic growth model mechanically coupled to the surrounding tissue. The resulting model was then used to estimate mechanical stress fields, which were then compared with edema visualized on FLAIR imaging using DICE similarity coefficients. DICE, model, and standard radiographic morphometric analysis parameters were evaluated using a receiver operating characteristic (ROC) curve for prediction of radiation necrosis or tumor progression. Multivariate logistic regression models were then constructed using mechanistic model parameters or advanced radiomic features. An independent validation was performed to evaluate predictive performance. Tumor cell proliferation rate resulted in ROC AUC=0.86, 95% CI: 0.76-0.95, P<0.0001, 74% sensitivity and 95% specificity) and DICE similarity coefficient associated with high stresses demonstrated an ROC AUC=0.93, 95% CI: 0.86-0.99, P<0.0001, 81% sensitivity and 95% specificity. In a multivariate logistic regression model using an independent validation dataset, mechanistic modeling parameters had an ROC AUC of 0.95, with 94% sensitivity and 96% specificity. Imaging-driven biophysical modeling of tumor growth represents a novel method for accurately predicting clinically significant tumor behavior.

Repeat-dose toxicity prediction with Generalized Read-Across (GenRA) using targeted transcriptomic data: A proof-of-concept case study

Read-across is a data gap filling technique utilized to predict the toxicity of a target chemical using data from similar analogues. Recent efforts such as Generalized Read-Across (GenRA) facilitate automated read-across predictions for untested chemicals. GenRA makes predictions of toxicity outcomes based on “neighboring” chemicals characterized by chemical and bioactivity fingerprints. Here we investigated the impact of biological similarities on neighborhood formation and read-across performance in predicting hazard (based on repeat-dose testing outcomes from US EPA ToxRefDB v2.0). We used targeted transcriptomic data on 93 genes for 1060 chemicals in HepaRG™ cells that measure nuclear receptor activation, xenobiotic metabolism, cellular stress, cell cycle progression, and apoptosis. Transcriptomic similarity between chemicals was calculated using binary hit-calls from concentration-response data for each gene. We evaluated GenRA performance in predicting ToxRefDB v2.0 hazard outcomes using the area under the Receiver Operating Characteristic (ROC) curve (AUC) for the baseline approach (chemical fingerprints) versus transcriptomic fingerprints and a combination of both (hybrid). For all endpoints, there were significant but only modest improvements in ROC AUC scores of 0.01 (2.1%) and 0.04 (7.3%) with transcriptomic and hybrid descriptors, respectively. However, for liver-specific toxicity endpoints, ROC AUC scores improved by 10% and 17% for transcriptomic and hybrid descriptors, respectively. Our findings suggest that using hybrid descriptors formed by combining chemical and targeted transcriptomic information can improve in vivo toxicity predictions in the right context.

Construction of a Prognostic Immune-Related LncRNA Risk Model for Lung Adenocarcinoma.

BackgroundLung adenocarcinoma (LUAD) originates mainly from the mucous epithelium and glandular epithelium of the bronchi. It is the most common pathologic subtype of non-small cell lung cancer (NSCLC). At present, there is still a lack of clear criteria to predict the efficacy of immunotherapy. The 5-year survival rate for LUAD patients remains low.MethodsAll data were downloaded from The Cancer Genome Atlas (TCGA) database. We used Gene Set Enrichment Analysis (GSEA) database to obtain immune-related mRNAs. Immune-related lncRNAs were acquired by using the correlation test of the immune-related genes with R version 3.6.3 (Pearson correlation coefficient cor = 0.5, P < 0.05). The TCGA-LUAD dataset was divided into the testing set and the training set randomly. Based on the training set to perform univariate and multivariate Cox regression analyses, we screened prognostic immune-related lncRNAs and given a risk score to each sample. Samples were divided into the high-risk group and the low-risk group according to the median risk score. By the combination of Kaplan–Meier (KM) survival curve, the receiver operating characteristic (ROC) (AUC) curve, the independent risk factor analysis, and the clinical data of the samples, we assessed the accuracy of the risk model. Gene Ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were performed on the differentially expressed mRNAs between the high-risk group and the low-risk group. The differentially expressed genes related to immune response between two risk groups were analyzed to evaluate the role of the model in predicting the efficacy and effects of immunotherapy. In order to explain the internal mechanism of the risk model in predicting the efficacy of immunotherapy, we analyzed the differentially expressed genes related to epithelial-mesenchymal transition (EMT) between two risk groups. We extracted RNA from normal bronchial epithelial cell and LUAD cells and verified the expression level of lncRNAs in the risk model by a quantitative real-time polymerase chain reaction (qRT-PCR) test. We compared our risk model with other published prognostic signatures with data from an independent cohort. We transfected LUAD cell with siRNA-LINC0253. Western blot analysis was performed to observed change of EMT-related marker in protein level.ResultsThrough univariate Cox regression analysis, 24 immune-related lncRNAs were found to be strongly associated with the survival of the TCGA-LUAD dataset. Utilizing multivariate Cox regression analysis, 10 lncRNAs were selected to establish the risk model. The K-M survival curves and the ROC (AUC) curves proved that the risk model has a fine predictive effect. The GO enrichment analysis indicated that the effect of the differentially expressed genes between high-risk and low-risk groups is mainly involved in immune response and intercellular interaction. The KEGG enrichment analysis indicated that the differentially expressed genes between high-risk and low-risk groups are mainly involved in endocytosis and the MAPK signaling pathway. The expression of genes related to the efficacy of immunotherapy was significantly different between the two groups. A qRT-PCR test verified the expression level of lncRNAs in LUAD cells in the risk model. The AUC of ROC of 5 years in the independent validation dataset showed that this model had superior accuracy. Western blot analysis verified the change of EMT-related marker in protein level.ConclusionThe immune lncRNA risk model established by us could better predict the prognosis of patients with LUAD.

Cardiac Operative Risk in Latin America: A Comparison of Machine Learning Models vs EuroSCORE-II

BackgroundMachine learning is a useful tool for predicting medical outcomes. This study aimed to develop a machine learning–based preoperative score to predict cardiac surgical operative mortality. MethodsWe developed various models to predict cardiac operative mortality using machine learning techniques and compared each model to European System for Cardiac Operative Risk Evaluation-II (EuroSCORE-II) using the area under the receiver operating characteristic (ROC) and precision-recall (PR) curves (ROC AUC and PR AUC) as performance metrics. The model calibration in our population was also reported with all models and in high-risk groups for gradient boosting and EuroSCORE-II. This study is a retrospective cohort based on a prospectively collected database from July 2008 to April 2018 from a single cardiac surgical center in Bogotá, Colombia. ResultsModel comparison consisted of hold-out validation: 80% of the data were used for model training, and the remaining 20% of the data were used to test each model and EuroSCORE-II. Operative mortality was 6.45% in the entire database and 6.59% in the test set. The performance metrics for the best machine learning model, gradient boosting (ROC: 0.755; PR: 0.292), were higher than those of EuroSCORE-II (ROC: 0.716, PR: 0.179), with a P value of .318 for the AUC of the ROC and .137 for the AUC of the PR. ConclusionsThe gradient boosting model was more precise than EuroSCORE-II in predicting mortality in our population based on ROC and PR analyses, although the difference was not statistically significant.

An evidence-based scoring system to diagnose acute rheumatic fever with carditis in children

BackgoundAcute rheumatic fever (ARF) with carditis and chronic rheumatic heart disease (RHD) may be difficult to differentiate and may lead to missed diagnosis of carditis. We aimed to determine differences between these groups and to develop a new scoring system for this purpose. MethodsChildren (N = 514; mean age 11.6 ± 2.8, range 5–18 years; 65% males) enrolled in a RHD registry at a tertiary care centre were studied. Clinical, laboratory and echocardiographic features of acute rheumatic carditis group (N = 126) were compared with chronic RHD group (N = 388). We performed multiple regression analysis and then developed a new scoring system using independent predictors. Accuracy of this scoring system was assessed using receiver operating characteristic (ROC) curve analysis. ResultsPatients with ARF and carditis were younger, more commonly had history of fever, arthritis/arthralgias and had worse NYHA class. On echocardiography, severity of mitral regurgitation, presence of mitral leaflet nodules, mitral valve prolapse and pericardial effusion were more common in ARF group. On multiple regression analysis, following features were found to be independently predictive of ARF with carditis: age -negative association, NYHA class, severity of mitral regurgitation, mitral leaflet nodules, erythrocyte sedimentation rate and antistreptolysin titer. Based on these 6 variables, a new scoring system (0−13) was developed. A score of 5 or more was found to be best threshold for diagnosis of carditis (ROC AUC 0.87, sensitivity 76%, specificity 79%). ConclusionA new scoring system based on independent statistical associations appears promising for differentiating ARF with carditis from chronic RHD.

Serum Metabonomic Study of Patients With Acute Coronary Syndrome Using Ultra-Performance Liquid Chromatography Orbitrap Mass Spectrometer

Acute coronary syndrome (ACS) can cause arrhythmia, heart failure, and even sudden death. Our aim in this study was to identify potential metabolic biomarkers in patients with ACS. The serum metabonomics approach based on ultra-performance liquid chromatography (UPLC)/Orbitrap mass spectrometer (MS) was used to analyze the serum samples from 45 patients with ACS and 29 healthy controls. Multivariate statistical analysis was used to screen for ACS biomarkers. In total, 69 biomarkers were identified to be enriched in 19 metabolic pathways; 43 biomarkers were significantly up-regulated, while 26 biomarkers were significantly down-regulated in the ACS group. The main classes were lyso-sphingolipid (SM), cinnamic acids, cholines, and primary amides. Receiver operating characteristic (ROC) curve analysis showed that lysoPC(20:4(8Z,11Z,14Z,17Z)/0:0) (ROC area under the curve, AUC = 0.936), SM(d18:0/16:0) (ROC AUC = 0.932), and SM(d18:1/14:0) (ROC AUC = 0.923) had a high ACS diagnostic ability. The AUC value of the diagnostic model constructed using these combined biomarkers was 0.96. Therefore, these biomarkers may improve the diagnostic efficacy of ACS. The findings of this study also implied that glycerophospholipid metabolism; the biosynthesis of unsaturated fatty acids; linoleic acid metabolism; and valine, leucine, and isoleucine biosynthesis played important roles in ACS. Network analysis by ingenuity pathway analysis (IPA) showed these biomarkers were correlated to the cardiac hypertrophy signaling pathway, ERK/MAPK signaling pathway, NF-kappa B signaling pathway, nitric oxide (NO) signaling pathway in cardiovascular system, and TLR-signaling pathway. These findings will help to improve the ability of accurate diagnosis and intervention of ACS.

Machine learning predicts live-birth occurrence before in-vitro fertilization treatment

In-vitro fertilization (IVF) is a popular method of resolving complications such as endometriosis, poor egg quality, a genetic disease of mother or father, problems with ovulation, antibody problems that harm sperm or eggs, the inability of sperm to penetrate or survive in the cervical mucus and low sperm counts, resulting human infertility. Nevertheless, IVF does not guarantee success in the fertilization. Choosing IVF is burdensome for the reason of high cost and uncertainty in the result. As the complications and fertilization factors are numerous in the IVF process, it is a cumbersome task for fertility doctors to give an accurate prediction of a successful birth. Artificial Intelligence (AI) has been employed in this study for predicting the live-birth occurrence. This work mainly focuses on making predictions of live-birth occurrence when an embryo forms from a couple and not a donor. Here, we compare various AI algorithms, including both classical Machine Learning, deep learning architecture, and an ensemble of algorithms on the publicly available dataset provided by Human Fertilisation and Embryology Authority (HFEA). Insights on data and metrics such as confusion matrices, F1-score, precision, recall, receiver operating characteristic (ROC) curves are demonstrated in the subsequent sections. The training process has two settings Without feature selection and With feature selection to train classifier models. Machine Learning, Deep learning, ensemble models classification paradigms have been trained in both settings. The Random Forest model achieves the highest F1-score of 76.49% in without feature selection setting. For the same model, the precision, recall, and area under the ROC Curve (ROC AUC) scores are 77%, 76%, and 84.60%, respectively. The success of the pregnancy depends on both male and female traits and living conditions. This study predicts a successful pregnancy through the clinically relevant parameters in In-vitro fertilization. Thus artificial intelligence plays a promising role in decision making process to support the diagnosis, prognosis, treatment etc.