China
Africa
Explore chapters and articles related to this topic
Reliable Diagnosis and Prognosis of COVID-19
Published in Varun Bajaj, G.R. Sinha, Computer-aided Design and Diagnosis Methods for Biomedical Applications, 2021
Marjan Mansourian, Hamid Reza Marateb, Maja von Cube, Sadaf Khademi, Mislav Jordanic, Miguel Ángel Mañanas, Harald Binder, Martin Wolkewitz
In epidemiology and medical data mining, each diagnostic accuracy measure has its importance [61]. Each provides essential information, and a missing accuracy index could be confusing to the readers and make it hard for detailed validation [62]. In binary classification problems, the single indices Se, Sp, PPV, NPV, and the composite indices F1-score [63], AUC, Matthews correlation coefficient (MCC) [64], accuracy, and diagnostic odds ratio (DOR) are usually provided [65]. In multi-class classification problems, on the other hand, proper performance indices such as macro-averaged F-score must be provided, in addition to the overall accuracy [66]. Regardless of the composite indices, Type I and Type II statistical errors must be provided in the diagnosis studies.
Quality Control for Automated Scoring in Large-Scale Assessment
Published in Duanli Yan, André A. Rupp, Peter W. Foltz, Handbook of Automated Scoring, 2020
Dan Shaw, Brad Bolender, Rick Meisner
We encourage practitioners to research or devise suitable metrics in the spirit of continuous improvement to help shed greater light on the precision of the scores in one’s data sets. For example, some of the most innovative assessment tasks, which might use different scales and rubrics than the traditional four- or six-point scales of direct writing assessment, require alternative metrics and standards to deal with their atypical score outputs. Scores produced on a dichotomous scale (‘0’ vs. ‘1’), for example, can benefit significantly from the application of evaluation metrics drawn from the broader fields of machine learning and medical science where binary diagnostics outcomes have a rich tradition of refined statistical analysis techniques. For example, the diagnostic odds ratio and Matthews correlation coefficient each provide additional useful information about the ratios of true positives / negatives to false positives / negatives, which is crucial information in dichotomous score outcomes.
Predictive validity of automated oscillometric blood pressure monitors for screening atrial fibrillation: a systematic review and meta-analysis
Published in Expert Review of Medical Devices, 2019
Seong-Hi Park, Kyung Ja June, Yun-Kyoung Choi
The meta-analysis was performed by using the MetaDiSc 1.4 software program [20]. For the analysis of pooled estimates in the diagnostic method meta-analysis, the use of a random effect model is recommended in order to reflect the heterogeneity between studies [16]. According to the general principles of statistical models, the pooled sensitivity and specificity, the positive and negative likelihood ratio, and the diagnostic odds ratio were analyzed by using a random effect model. In addition, for the sROC statistics, the area under the curve (AUC) and the index Q* value were analyzed in order to describe the test accuracy. With regard to the AUC value, a test of AUC = 0.5 was considered as a noninformative test, 0.5 < AUC ≤ 0.7 as a less accurate test, 0.7 < AUC ≤ 0.9 as a moderately accurate test, 0.9 < AUC < 1 as a highly accurate test, and AUC = 1 as a perfect test [21]. For the index Q* value, representing the point where the sensitivity and specificity are equal in the ROC curve, a value of 1 was considered to represent 100% accuracy [22]. The heterogeneity of the reports was evaluated by performing the Higgins I2 test at a significance level less than 5%. The I2 value was interpreted as representing a low heterogeneity for I2 ≤ 25%, a moderate heterogeneity for 25% < I2 ≤ 75%, and a high heterogeneity for I2 > 75% [23]. For the analysis of the heterogeneity between the reports, sub-group analyses were performed according to the type of AOBPM, reference standard test, and the average age of the subjects.
Diagnostic accuracy of liver stiffness on two-dimensional shear wave elastography for detecting clinically significant portal hypertension: a meta-analysis
Published in Expert Review of Medical Devices, 2023
Bingtian Dong, Yuping Chen, Yongjian Chen, Huaming Wang, Guorong Lyu
We extracted the data from the included studies, and 2 × 2 tables were then constructed for further analysis. In order to examine the accuracy of 2D-SWE for diagnosing CSPH, three measures of diagnostic test performance were used in this meta-analysis, including the summary AUROC, the summary diagnostic odds ratio (DOR), and the summary sensitivity and specificity. The summary receiver operating characteristic (SROC) curve was constructed using the data from the included studies to calculate the summary AUROC of 2D-SWE for detecting CSPH. Further, a Der-Simonian and Laird random-effects model was used to calculate the summary DOR. The bivariate meta-analytic approach was used to calculate the summary sensitivity and specificity.
The role of ultrasonography in the diagnosis of anterior cruciate ligament injury: A systematic review and meta-analysis
Published in European Journal of Sport Science, 2018
Jianhong Wang, Huaiyu Wu, Fajin Dong, Binbin Li, Zhanghong Wei, Quanzhou Peng, Duo Dong, Min Li, Jinfeng Xu
We used fourfold tables to record the data as true positive, true negative, false positive, false negative. Statistical analyses were performed using RevMan 5.3 and Stata 12.0. Rev. 5.3 was used to assess the methodological quality of the eligible studies. Stata was used to pool statistical indexes and draw statistical graphs such as the pool of forest graph of LR+, LR−, and diagnostic odds ratio (DOR) with corresponding 95% confidence intervals (CI), and the area under summary receiver operating characteristic curve (SROC).