Explore the vast range of titles available.

To evaluate the first experience of real-time instantaneous wave–free ratio (iFR) measurement by clinicians. The iFR is a new vasodilator-free index of coronary stenosis severity, calculated as a trans-lesion pressure ratio during a specific period of baseline diastole, when distal resistance is lowest and stable. Because all previous studies have calculated iFR offline, the feasibility of real-time iFR measurement has never been assessed. Three hundred ninety-two stenoses with angiographically intermediate stenoses were included in this multicenter international analysis. Instantaneous wave–free ratio and fractional flow reserve (FFR) were performed in real time on commercially available consoles. The classification agreement of coronary stenoses between iFR and FFR was calculated. Instantaneous wave–free ratio and FFR maintain a close level of diagnostic agreement when both are measured by clinicians in real time (for a clinical 0.80 FFR cutoff: area under the receiver operating characteristic curve [ROCAUC] 0.87, classification match 80%, and optimal iFR cutoff 0.90; for a ischemic 0.75 FFR cutoff: iFR ROCAUC 0.90, classification match 88%, and optimal iFR cutoff 0.85; if the FFR 0.75-0.80 gray zone is accounted for: ROCAUC 0.93, classification match 92%). When iFR and FFR are evaluated together in a hybrid decision-making strategy, 61% of the population is spared from vasodilator while maintaining a 94% overall agreement with FFR lesion classification. When measured in real time, iFR maintains the close relationship to FFR reported in offline studies. These findings confirm the feasibility and reliability of real-time iFR calculation by clinicians. [Display omitted]

Compare the area under the receiver operating characteristic curve (AUC) vs. the area under the precision–recall curve (AUPRC) in summarizing the performance of a diagnostic biomarker according to the disease prevalence. A simulation study was performed considering different sizes of diseased and nondiseased groups. Values of a biomarker were sampled with various variances and differences in mean values between the two groups. The AUCs and the AUPRCs were examined regarding their agreement and vs. the positive predictive value (PPV) and the negative predictive value (NPV) of the biomarker. With a disease prevalence of 50%, the AUC and the AUPRC showed high correlations with the PPV and the NPV (ρ > 0.95). With a prevalence of 1%, small PPV and AUPRC values (<0.2) but high AUC values (>0.9) were found. The AUPRC reflected better than the AUC the discriminant ability of the biomarker; it had a higher correlation with the PPV (ρ = 0.995 vs. 0.724; P < 0.001). In uncommon and rare diseases, the AUPRC should be preferred to the AUC because it summarizes better the performance of a biomarker.

The 1-minute sit-to-stand (STS) test could be valuable to assess the level of exercise tolerance in chronic obstructive pulmonary disease (COPD). There is a need to provide the minimal important difference (MID) of this test in pulmonary rehabilitation (PR). COPD patients undergoing the 1-minute STS test before PR were included. The test was performed at baseline and the end of PR, as well as the 6-minute walk test, and the quadriceps maximum voluntary contraction (QMVC). Home and community-based programs were conducted as recommended. Responsiveness to PR was determined by the difference in the 1-minute STS test between baseline and the end of PR. The MID was evaluated using distribution and anchor-based methods. Forty-eight COPD patients were included. At baseline, the significant predictors of the number of 1-minute STS repetitions were the 6-minute walk distance (6MWD) ( =0.574; <10 ), age ( =-0.453; =0.001), being on long-term oxygen treatment ( =-0.454; =0.017), and the QMVC ( =0.424; =0.031). The multivariate analysis explained 75.8% of the variance of 1-minute STS repetitions. The improvement of the 1-minute STS repetitions at the end of PR was 3.8±4.2 ( <10 ). It was mainly correlated with the change in QMVC ( =0.572; =0.004) and 6MWD ( =0.428; =0.006). Using the distribution-based analysis, an MID of 1.9 (standard error of measurement method) or 3.1 (standard deviation method) was found. With the 6MWD as anchor, the receiver operating characteristic curve identified the MID for the change in 1-minute STS repetitions at 2.5 (sensibility: 80%, specificity: 60%) with area under curve of 0.716. The 1-minute STS test is simple and sensitive to measure the efficiency of PR. An improvement of at least three repetitions is consistent with physical benefits after PR.

Background In classification and diagnostic testing, the receiver-operator characteristic (ROC) plot and the area under the ROC curve (AUC) describe how an adjustable threshold causes changes in two types of error: false positives and false negatives. Only part of the ROC curve and AUC are informative however when they are used with imbalanced data. Hence, alternatives to the AUC have been proposed, such as the partial AUC and the area under the precision-recall curve. However, these alternatives cannot be as fully interpreted as the AUC, in part because they ignore some information about actual negatives. Methods We derive and propose a new concordant partial AUC and a new partial c statistic for ROC data—as foundational measures and methods to help understand and explain parts of the ROC plot and AUC. Our partial measures are continuous and discrete versions of the same measure, are derived from the AUC and c statistic respectively, are validated as equal to each other, and validated as equal in summation to whole measures where expected. Our partial measures are tested for validity on a classic ROC example from Fawcett, a variation thereof, and two real-life benchmark data sets in breast cancer: the Wisconsin and Ljubljana data sets. Interpretation of an example is then provided. Results Results show the expected equalities between our new partial measures and the existing whole measures. The example interpretation illustrates the need for our newly derived partial measures. Conclusions The concordant partial area under the ROC curve was proposed and unlike previous partial measure alternatives, it maintains the characteristics of the AUC. The first partial c statistic for ROC plots was also proposed as an unbiased interpretation for part of an ROC curve. The expected equalities among and between our newly derived partial measures and their existing full measure counterparts are confirmed. These measures may be used with any data set but this paper focuses on imbalanced data with low prevalence. Future work Future work with our proposed measures may: demonstrate their value for imbalanced data with high prevalence, compare them to other measures not based on areas; and combine them with other ROC measures and techniques.

To evaluate how often newly developed risk prediction models undergo external validation and how well they perform in such validations. We reviewed derivation studies of newly proposed risk models and their subsequent external validations. Study characteristics, outcome(s), and models' discriminatory performance [area under the curve, (AUC)] in derivation and validation studies were extracted. We estimated the probability of having a validation, change in discriminatory performance with more stringent external validation by overlapping or different authors compared to the derivation estimates. We evaluated 127 new prediction models. Of those, for 32 models (25%), at least an external validation study was identified; in 22 models (17%), the validation had been done by entirely different authors. The probability of having an external validation by different authors within 5 years was 16%. AUC estimates significantly decreased during external validation vs. the derivation study [median AUC change: −0.05 (P < 0.001) overall; −0.04 (P = 0.009) for validation by overlapping authors; −0.05 (P < 0.001) for validation by different authors]. On external validation, AUC decreased by at least 0.03 in 19 models and never increased by at least 0.03 (P < 0.001). External independent validation of predictive models in different studies is uncommon. Predictive performance may worsen substantially on external validation.

Background The random forest (RF) method is a commonly used tool for classification with high dimensional data as well as for ranking candidate predictors based on the so-called random forest variable importance measures (VIMs). However the classification performance of RF is known to be suboptimal in case of strongly unbalanced data, i.e. data where response class sizes differ considerably. Suggestions were made to obtain better classification performance based either on sampling procedures or on cost sensitivity analyses. However to our knowledge the performance of the VIMs has not yet been examined in the case of unbalanced response classes. In this paper we explore the performance of the permutation VIM for unbalanced data settings and introduce an alternative permutation VIM based on the area under the curve (AUC) that is expected to be more robust towards class imbalance. Results We investigated the performance of the standard permutation VIM and of our novel AUC-based permutation VIM for different class imbalance levels using simulated data and real data. The results suggest that the new AUC-based permutation VIM outperforms the standard permutation VIM for unbalanced data settings while both permutation VIMs have equal performance for balanced data settings. Conclusions The standard permutation VIM loses its ability to discriminate between associated predictors and predictors not associated with the response for increasing class imbalance. It is outperformed by our new AUC-based permutation VIM for unbalanced data settings, while the performance of both VIMs is very similar in the case of balanced classes. The new AUC-based VIM is implemented in the R package party for the unbiased RF variant based on conditional inference trees. The codes implementing our study are available from the companion website: http://www.ibe.med.uni-muenchen.de/organisation/mitarbeiter/070_drittmittel/janitza/index.html

Several machine learning (ML) algorithms have been increasingly utilized for cardiovascular disease prediction. We aim to assess and summarize the overall predictive ability of ML algorithms in cardiovascular diseases. A comprehensive search strategy was designed and executed within the MEDLINE, Embase, and Scopus databases from database inception through March 15, 2019. The primary outcome was a composite of the predictive ability of ML algorithms of coronary artery disease, heart failure, stroke, and cardiac arrhythmias. Of 344 total studies identified, 103 cohorts, with a total of 3,377,318 individuals, met our inclusion criteria. For the prediction of coronary artery disease, boosting algorithms had a pooled area under the curve (AUC) of 0.88 (95% CI 0.84–0.91), and custom-built algorithms had a pooled AUC of 0.93 (95% CI 0.85–0.97). For the prediction of stroke, support vector machine (SVM) algorithms had a pooled AUC of 0.92 (95% CI 0.81–0.97), boosting algorithms had a pooled AUC of 0.91 (95% CI 0.81–0.96), and convolutional neural network (CNN) algorithms had a pooled AUC of 0.90 (95% CI 0.83–0.95). Although inadequate studies for each algorithm for meta-analytic methodology for both heart failure and cardiac arrhythmias because the confidence intervals overlap between different methods, showing no difference, SVM may outperform other algorithms in these areas. The predictive ability of ML algorithms in cardiovascular diseases is promising, particularly SVM and boosting algorithms. However, there is heterogeneity among ML algorithms in terms of multiple parameters. This information may assist clinicians in how to interpret data and implement optimal algorithms for their dataset.

Non-invasive fibrosis scores are widely used to identify/exclude advanced fibrosis in patients with non-alcoholic fatty liver disease (NAFLD). However, these scores were principally developed and validated in patients aged between 35 and 65 years of age. The objective of this study was to assess the effect of age on the performance of non-invasive fibrosis tests in NAFLD. Patients were recruited from European specialist hepatology clinics. The cohort was divided into five age-based groups: ≤35 (n=74), 36-45 (n=96), 46-55 (n=197), 56-64 (n=191), and ≥65 years (n=76), and the performance of the aspartate aminotransferase (AST)/alanine transaminase (ALT) ratio, fibrosis 4 (FIB-4), and NAFLD fibrosis score (NFS) for advanced fibrosis (stage F3-F4) for each group was assessed using liver biopsy as the standard. Six hundred and thirty-four patients were included. The diagnostic accuracy of the AST/ALT ratio was lower than NFS and FIB-4 in all the age groups. The AST/ALT ratio, NFS, and FIB-4 score performed poorly for a diagnosis of advanced fibrosis in those aged ≤35 years (area under the receiver operating characteristic curves (AUROCs 0.52, 0.52, and 0.60, respectively). For all groups >35 years, AUROCs for advanced fibrosis were similar for the NFS and FIB-4 score (range 0.77-0.84). However, the specificity for advanced fibrosis using the FIB-4 and NFS declined with age, becoming unacceptably low in those aged ≥65 years (35% for FIB-4 and 20% for NFS). New cutoffs were derived (and validated) for those aged ≥65 years, which improved specificity to 70% without adversely affecting sensitivity (FIB-4 2.0, sensitivity 77%; NFS 0.12, sensitivity 80%). The NFS and FIB-4 scores have similar accuracy for advanced fibrosis in patients aged >35 years. However, the specificity for advanced fibrosis is unacceptably low in patients aged ≥65 years, resulting in a high false positive rate. New thresholds for use in patients aged ≥65 years are proposed to address this issue.

Traditional fact checking by expert journalists cannot keep up with the enormous volume of information that is now generated online. Computational fact checking may significantly enhance our ability to evaluate the veracity of dubious information. Here we show that the complexities of human fact checking can be approximated quite well by finding the shortest path between concept nodes under properly defined semantic proximity metrics on knowledge graphs. Framed as a network problem this approach is feasible with efficient computational techniques. We evaluate this approach by examining tens of thousands of claims related to history, entertainment, geography, and biographical information using a public knowledge graph extracted from Wikipedia. Statements independently known to be true consistently receive higher support via our method than do false ones. These findings represent a significant step toward scalable computational fact-checking methods that may one day mitigate the spread of harmful misinformation.

A deep-learning system that was applied to 14,341 fundus photographs differentiated optic disks with papilledema from normal disks with 96.4% sensitivity and 84.7% specificity in an external-testing data set. The prevalence of papilledema was 9.5%, yielding positive and negative predictive values of 39.8% and 99.6%, respectively.