Joint Robust Linear Regression and Anomaly Detection in Poisson noise using Expectation-Propagation

Author(s):  
D. Yao ◽  
Y. Altmann ◽  
S. McLaughlin ◽  
M. E. Davies
2019 ◽  
Vol 106 ◽  
pp. 65-70 ◽  
Author(s):  
Jie Liu ◽  
Sheng Li ◽  
Faezeh Jahanmiri-Nezhad ◽  
William Zev Rymer ◽  
Ping Zhou

Injury ◽  
2020 ◽  
Author(s):  
Epaminondas Markos Valsamis ◽  
Henry Husband ◽  
Daniel Burchette ◽  
Milan Milošević ◽  
Bore Bakota

2011 ◽  
Vol 27 (6) ◽  
pp. 815-821 ◽  
Author(s):  
V. M. Lourenço ◽  
A. M. Pires ◽  
M. Kirst

Blood ◽  
2014 ◽  
Vol 124 (21) ◽  
pp. 2717-2717 ◽  
Author(s):  
Cody Cichowitz ◽  
C. Patrick Carroll ◽  
John J. Strouse ◽  
Carlton Haywood ◽  
Sophie Lanzkron

Abstract Introduction: Studies have described neuroimaging abnormalities and neurocognitive dysfunction in adults living with sickle cell anemia and no previous history of neurological impairment. At the Johns Hopkins Sickle Cell Center for Adults, we have been administering the Montreal Cognitive Assessment (MoCA) at regularly scheduled outpatient appointments as part of routine screening and clinical care. The MoCA is scored out of 30 points and consists of 28 questions grouped into seven domains of cognitive function: visual-spatial and executive, naming, attention, language, abstraction, delayed recall and orientation. While the MoCA has yet to be validated as a screening tool for cognitive impairment in adults with sickle cell disease (SCD), it has been widely used and validated in other populations to screen for mild cognitive impairment with the cutoff typically ranging from 22 to 26. The objective of this study was to describe the results of MoCA testing in a sample of adults with SCD and to explore predictors of MoCA performance using data from a retrospective chart review. Methods: A cross-sectional study was completed of the first 100 MoCAs administered to adult patients with SCD. Demographic, laboratory and clinical data were collected from each participant’s medical record up to the date that the MoCA was administered. The internal validity of each MoCA domain was analyzed using standard psychometric statistics, including a Cronbach-α score and factor analysis. Bivariate analysis was completed using Mann–Whitney U tests and Spearman Rank correlations. We identified independent predictors of MoCA performance using a multivariable robust linear regression. Age, hemoglobin and genotype were included in the multivariable analysis along with any variable found to have an association with MoCA score (p-values ≤ .10). Results: The distribution of scores is displayed in Figure 1; the mean score was 24.5 with a standard deviation of 4.1. The visual-spatial and executive function and attention domains showed strong correlation with overall test performance and demonstrated high measures of internal validity. Education, gender, weight, aspartate aminotransferase, cerebral vascular accident (CVA), chronic kidney disease (CKD) and a history of hydroxyurea therapy were associated with MoCA scores in bivariate analysis. The results of multivariable analysis are displayed in Table 1. Education was found to be a significant independent predictor of increased MoCA score, while CVA and CKD were found to be significant predictors of decreased MoCA score. When limited to the 64 participants with SS or Sβ0 Thalassemia, education and a history of hydroxyurea therapy were found to be significant independent predictors of increased score, while CKD was found to be a significant predictor of decreased score. Conclusion: A screening tool for neurocognitive dysfunction in adults with SCD is needed in order to identify those that require more definitive testing. The significant association of MoCA score with both education and CVA supports the potential validity of this measure as a screening tool in this patient population. Further validation of this tool is needed as well as an exploration into the possible relationship between improved MoCA performance and hydroxyurea use. Figure 1 Figure 1. Table 1: Multivariable Analysis Model 1: Robust Linear Regression for MoCA Score (n = 89) Independent Variables Coefficient [95% Confidence Interval] P-value Education – Completed > 12th Grade 3.1 1.5 4.7 .0003 Gender - Male 1.4 -0.13 2.8 .0738 Age (years) -0.043 -0.11 0.024 .2055 Genotype – SS or Sβ0 Thalassemia 1.3 -1.3 3.9 .3350 Hemoglobin (g/dL) 0.033 -0.62 0.69 .9196 Weight (lbs) 0.014 -0.0052 0.034 .1504 Aspartate Aminotransferase (Units/L) -0.035 -0.078 0.0090 .1179 CVA -3.3 -5.7 -0.90 .0079 CKD -3.1 -6.2 -0.056 .0460 Model 2: Robust Linear Regression for MoCA Score for participants with SS or Sβ0 Thalassemia (n = 64) Independent Variables Coefficient [95% Confidence Interval] P-value Education – Completed > 12th Grade 3.7 2.2 5.2 .0000 Gender - Male 0.95 -0.71 2.6 .2561 Age (years) -0.043 -0.13 0.046 .3363 Hemoglobin (g/dL) -0.010 -0.69 0.67 .9770 Weight (lbs) 0.0067 -0.015 0.029 .5451 Aspartate Aminotransferase (Units/L) -0.044 -0.089 0.00062 .0531 CVA -1.9 -4.7 0.92 .1832 CKD -3.4 -6.4 -0.27 .0334 History of Hydroxyurea Therapy 2.1 0.14 4.0 .0364 Disclosures Haywood: NHLBI: Research Funding. Lanzkron:NHLBI: Research Funding.


Sign in / Sign up

Export Citation Format

Share Document