China 
Africa 
Explore chapters and articles related to this topic
Resources and datasets for radiomics
Published in Ruijiang Li, Lei Xing, Sandy Napel, Daniel L. Rubin, Radiomics and Radiogenomics, 2019
Ken Chang, Andrew Beers, James Brown, Jayashree Kalpathy-Cramer
The differences in methods of feature extraction have made the radiomic features difficult to reproduce.10,29–33 Furthermore, certain features are sensitive to segmentation. Kalpathy-Cramer et al. found that 32% of features had a concordance correlation coefficient of <0.75 with different underlying segmentations.10 Dercele et al. found that the entropy imaging feature within the tumor region is not just correlated with tumor type, but also with tumor size which is a major confounding factor.34 The authors found underestimation of entropy was greater in smaller region of interest. As such, they recommend a minimum region of interest size for accurate reproduction of entropy. Dercele et al. also found that entropy of the tumor region was correlated with entropy of reference normal tissue, which may also confound interpretation of the feature.34 Choice of acquisition parameters and CT scanner model can also have significant effect on radiomic features.29,30 Specifically, Berenguer et al. found that only 43.1% and 89.3% of features were reproducible when pitch factor and reconstruction kernels were varied, respectively.31 Furthermore, when the model of CT scanner was varied, the proportion of reproducible features ranged from 15.8% to 85.3%, depending on the phantom material.31 For example, Mackin et al. found that X-ray tube current levels had little effect on radiomic features extracted from phantoms with tissue-like textures on CT imaging.35
Using Digital Health Technology to Promote Cardiovascular Disease Risk Reduction in Secondary Prevention
Published in James M. Rippe, Lifestyle Medicine, 2019
Neil F. Gordon, Richard D. Salmon, Mandy K. Salmon, Prabakar Ponnusamy
From a wearable technology perspective, the utility of wrist-worn heart rate monitors is of particular relevance to home-based cardiac rehabilitation programming and exercise prescription. Wang et al. recently compared the accuracy of four popular wrist-worn heart rate monitors (namely, Fitbit Charge HR, Apple Watch, Mio Alpha, and Basis Peak) to that of standard electrocardiographic limb leads and a Polar H7 chest strap monitor in 50 healthy adults.23 Heart rate was assessed at rest and at treadmill speeds of two, three, four, five, and six miles per hour. Participants exercised at each speed for three minutes to achieve a steady state and heart rate was recorded instantaneously at the three-minute point. After completion of the treadmill protocol, heart rate was recorded at 30, 60, and 90 seconds of recovery. When compared with electrocardiographically measured heart rates, the heart rate monitors had variable accuracy as assessed by the concordance correlation coefficient. Of the four wrist-worn devices, the Apple Watch and Mio Fuse both had the best accuracy with concordance correlation coefficients of 0.91 whereas the Fitbit Charge HR and Basis Peak had concordance correlation coefficients of 0.84 and 0.83, respectively. None of the wrist-worn devices were as accurate as the Polar H7 chest strap monitor, which had a concordance correlation coefficient of 0.99. In general, the accuracy of the wrist-worn monitors was best at rest and diminished during exercise. While the Basis Peak overestimated heart rate during moderate exercise, with median differences of –8.9 and –7.3 beats/minute at 2 (p < .001) and three miles/hour (p = .001), respectively, the Fitbit Charge HR underestimated heart rate during more vigorous exercise, with median differences of 7.2 and 6.4 beats/minute at four (p < .001) and six miles/hour (p < .001), respectively (comparing each device with electrocardiogram). Analyses further showed that variability occurred across the spectrum of midrange heart rates during exercise, with less variability at the tail ends. The Apple Watch and Mio Fuse had 95% of differences fall within –27 and +29 beats/minute of the electrocardiogram, while Fitbit Charge HR had 95% of values within –34 and +39 beats/minute, and the corresponding values for the Basis Peak were within –39 and +33 beats/minute. The investigators concluded that electrode-containing chest monitors should be used when accurate heart rate measurement is imperative. They further recommended that, because cardiac patients increasingly rely on heart rate monitors to stay within prescribed target heart rate ranges, appropriate validation of these devices in such patients is imperative.
Concordance Analysis
Published in Atanu Bhattacharjee, Bayesian Approaches in Oncology Using R and OpenBUGS, 2020
It is always required for new diagnosis procedure to diagnose cancer more accurately. Now the challenge is to establish the accuracy of the new diagnostic tool in presences of the existing one. The concordance correlation coefficient is useful to quantify the agreement between two raters measured independently on the same subject. A new diagnostic tool promotion becomes challenging in the presence of an available tool. Now new tool needs to proof as authoritative or as equal in the presence of diagnostic tool. Now the different diagnostic tool is required to be compared to explore the agreement on measurement. Unless the new diagnostic tools are performed equal or better, we can not use in regular news. In this context, the concordance correlation coefficient becomes useful. There are different tools available on continuous data like the coefficient of variation, t-test, intra-class correlation, Pearson correlation coefficient, and least square analysis. However, these are not useful to work toward agreement analysis. There are a few limitations. The Intra-class correlation coefficient assumed that the readings between two observers are interchangeable. It is not robust enough. Now the least-squares analysis is not suitable enough for the null hypothesis in the presence of a residual error that is very large or small [144]. Similarly, the Pearson correlation is also not enough in this context. Now the concordance correlation coefficient (CCC) supports for differentiating the concordance measurement [145]. The computation of the Z-transformation is difficult. The methodological work supports that CCC provides similar to the value of the Pearson correlation coefficient if the mean and variance of the measurement of interest [144]. There is an illustration of the CCC [146]. It is based on the covariance-based index. Now the stratified CCC [147] and the generalized estimating equation is established [148]. The size of the tumor is mainly concerned in any regular or experimental cancer therapy. Determination of tumor size by diagnostic procedures is an important key indicator for any therapeutic success. Recently, several types of diagnostic procedures are available for tumor size detection like computed tomography (CT) and magnetic resonance imaging (MRI) with advanced technology. Generally, pre and post-therapy tumor size provide therapeutic results in any specific direction. It is ideal that the same diagnostic procedure should be applied to detect the tumor size before and after a therapeutic effect to reduce the diagnostic testing variation. Simultaneously, same way interpretation about tumor size needs to be performed. Broadly, two types of approaches are available for detection of tumor size, either through tumor volume or by the maximum area covered by the tumor. This chapter is about proposing the Bayesian counterpart to compute CCC for continuous data.
From monocular photograph to angle lambda: A new clinical approach for quantitative assessment
Published in Journal of Binocular Vision and Ocular Motility, 2022
Maxence Rateaux, Dominique Bremond-Gignac, Matthieu P. Robert
Lin’s concordance correlation coefficient was equal to 0.99 (confidence interval = [0.96; 0.99]). No significant difference was found between the values of angle λ calculated with measured values and those calculated with theoretical values (series of first pictures: p = 0.3017; series of repeated pictures: p = 0.2910). There was no significant difference between the measured corneal diameter and the theoretical value (11.71 mm), while there was a significant difference between the DAC and the theoretical value (3.4 mm) (respectively, p = 0.147 and p = 0.0021). The mean angle λ approximated by Pentacam and law of cosines was +2.70° ± 3.18°. The Bland–Altman plot for the 20 eyes is shown in Figure 5. Both methods provided comparable results, with a systematic deviation of −0.071°, a 95% confidence interval of –1.86 to +1.79, and no significant difference between the average of both measurements (p = 0.671). Then, the linear equation between λpentacam and λphotograph was R2 = 0.9229.
Unattended compared to traditional blood pressure measurement in patients with rheumatoid arthritis: a randomised cross-over study
Published in Annals of Medicine, 2021
Elena Bartoloni, Fabio Angeli, Elisa Marcucci, Carlo Perricone, Giacomo Cafaro, Clara Riccini, Lorenzo Spighi, Benedetta Gildoni, Claudio Cavallini, Paolo Verdecchia, Roberto Gerli
The strength of the relations between variables was assessed by partial correlation analysis [29]. Differences between two dependent correlation coefficients were evaluated according to established methods [30]. The concordance correlation coefficient was also used to compute a measurement of precision [31]. The following descriptive scale for values of the concordance correlation coefficient (for continuous variables) was employed to evaluate strength of agreement: <0.90, 0.90–0.95, and 0.95–0.99 for poor, moderate, and substantial strength of agreement, respectively [32]. Being a cross-over design experiment, we used analysis of variance (ANOVA) for a 2 × 2 crossover study to compute estimation of type of BP measurements and sequence effects [33]. In 2-tailed tests, p values <.05 were considered statistically significant.
Reliability and validity of outcome measures used for urinary incontinence in patients with stroke: a narrative review
Published in Physical Therapy Reviews, 2020
Emma Doyle, Jake Brettkelly, Rebecca Buhler, Tim Lovett, Luke O’Neil, Daniela Aldabe
For the purpose of this study, reliability findings were interpreted using categories. Studies that reported Intraclass Correlation (ICC) values were categorized as: <0.40 poor reliability, 0.40–0.75 fair to good reliability, and >0.75 excellent reliability [17]. Kappa values were interpreted as: <0.59 poor reliability, 0.60 to 0.79 fair reliability, 0.80 to 0.90 good reliability, and >0.90 excellent reliability [18]. Concordance Correlation Coefficient (CCC) values were defined as: <0.50 poor correlation, 0.50 to 0.70 fair correlation, 0.70 to 0.90 good correlation, and >0.90 excellent correlation [19]. Finally, Spearman’s Correlation Coefficient (r s) values were interpreted as: −0.3 to 0.3 = poor correlation, −0.6 to −0.3, and 0.3 to 0.6 = fair correlation, −0.8 to −0.6, and 0.6 to 0.8 = good correlation, and <-0.8, and >0.8 = excellent correlation [20].