Validity of Screening Audiological Tests in Detecting Signs of Ototoxicity in Cystic Fibrosis Children - Dissertation Example

?Results Simple Linear regression equation was plotted and correlation co-efficients were analyzed along with kappa values. Result1: Left Standard PTA (Y) to Left Screening HFPTA(X) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 0.6 S.D. (beta) 0.0974 T-Stat (beta) 6.1619 P-Value (beta) 0 Correlation 0.7474 R-squared 0.5586 Result 2:Right Standard PTA (Y) to Right Screening HFPTA(X) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 1 S.D. (beta) 0 T-Stat (beta) 44157213008362808 P-Value (beta) 0 Correlation 1 R-squared 1 Result3 : Right Screening HFPTA(X) to Right Diagnostic HFPTA(Y) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 1 S.D. (beta) 0 T-Stat (beta) 44157213008362808 P-Value (beta) 0 Correlation 1 R-squared 1 Result 4: Left Screening HFPTA to Left Diagnostic HFPTA Y = alpha + beta X Intercept (alpha) 0.037 Slope (beta) 0.563 S.D. (beta) 0.1307 T-Stat (beta) 4.3063 P-Value (beta) 2e-04 Correlation 0.6181 R-squared 0.382 Result 5: Right Screening HFPTA (X) to Right Diagnostic DPOAE(Y) Y = alpha + beta X Intercept (alpha) 0.037 Slope (beta) 0.563 S.D. (beta) 0.1307 T-Stat (beta) 4.3063 P-Value (beta) 2e-04 Correlation 0.6181 R-squared 0.382 Result 6: Left Screening HFPTA (X) to Left Diagnostic DPOAE(Y) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 0.6 S.D. (beta) 0.0974 T-Stat (beta) 6.1619 P-Value (beta) 0 Correlation 0.7474 R-squared 0.5586 Result 7:Right Screening DPOAE (X) to Right Screening HFPTA(Y) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 0.75 S.D. (beta) 0.0845 T-Stat (beta) 8.8741 P-Value (beta) 0 Correlation 0.851 R-squared 0.7241 Result 8: Left Screening DPOAE(X) to Left Screening HFPTA(Y) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 1 S.D. (beta) 0 T-Stat (beta) 15129861409362786 P-Value (beta) 0 Correlation 1 R-squared 1 Result9: Left Standard PTA(Y) to Left Screening DPOAE(X) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 1 S.D. (beta) 0 T-Stat (beta) 15129861409362786 P-Value (beta) 0 Correlation 1 R-squared 1 Result 10: Right Standard PTA(Y) to Right Screening DPOAE(X) Y = alpha + beta X Intercept (alpha) 0.0714 Slope (beta) 0.6786 S.D. (beta) 0.1576 T-Stat (beta) 4.3063 P-Value (beta) 2e-04 Correlation 0.6181 R-squared 0.382 Result 11: Left Screening DPOAE(X) to Left Diagnostic HFPTA(Y) Y = alpha + beta X Intercept (alpha) 0.069 Slope (beta) 0.931 S.D. (beta) 0.1511 T-Stat (beta) 6.1619 P-Value (beta) 0 Correlation 0.7474 R-squared 0.5586 Result12: Right Screening DPOAE(X) to Right Diagnostic HFPTA(Y) Y = alpha + beta X Intercept (alpha) 0.0345 Slope (beta) 0.9655 S.D. (beta) 0.1088 T-Stat (beta) 8.8741 P-Value (beta) 0 Correlation 0.851 R-squared 0.7241 Result 13: Right Screening DPOAE(X) to Right Diagnostic DPOAE(Y) Y = alpha + beta X Intercept (alpha) 0 Slope (beta) 1 S.D. (beta) 0 T-Stat (beta) 17273101793064750 P-Value (beta) 0 Correlation 1 R-squared 1 Result14: Left Screening DPOAE(X) to Left Diagnostic DPOAE(Y) Y = alpha + beta X Intercept (alpha) 0.04 Slope (beta) 0.96 S.D. (beta) 0.0765 T-Stat (beta) 12.5499 P-Value (beta) 0 Correlation 0.9165 R-squared 0.84 Result 15: Overall Screening(X) To Diagnostic Tests(Y) A B Total A 2927 0529 3456 B 0327 0305 632 Total 3254 834 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2879.9 ( 70.45% of the observations) Kappa= 0.291 SE of kappa = 0.018 95% confidence interval: From 0.255 to 0.327 The strength of agreement is considered to be 'fair'. Results: Right Standard PTA to Right Screening HFPTA A B Total A 2927 0529 3456 B 0327 0305 632 Total 3254 834 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2879.9 ( 70.45% of the observations) Kappa= 0.291 SE of kappa = 0.018 95% confidence interval: From 0.255 to 0.327 The strength of agreement is considered to be 'fair'. Results: Left Standard PTA to Left Screening HFPTA A B Total A 2727 0527 3254 B 0527 0505 1032 Total 3254 1032 4286 Number of observed agreements: 3232 ( 75.41% of the observations) Number of agreements expected by chance: 2719.0 ( 63.44% of the observations) Kappa= 0.327 SE of kappa = 0.016 95% confidence interval: From 0.295 to 0.360 The strength of agreement is considered to be 'fair'. Right Screening HFPTA to Right Diagnostic HFPTA A B Total A 2727 0527 3254 B 0527 0505 1032 Total 3254 1032 4286 Number of observed agreements: 3232 ( 75.41% of the observations) Number of agreements expected by chance: 2719.0 ( 63.44% of the observations) Kappa= 0.327 SE of kappa = 0.016 95% confidence interval: From 0.295 to 0.360 The strength of agreement is considered to be 'fair'. Left Screening HFPTA to Left Diagnostic HFPTA A B Total A 2728 0427 3155 B 0528 0504 1032 Total 3256 931 4187 Number of observed agreements: 3232 ( 77.19% of the observations) Number of agreements expected by chance: 2682.9 ( 64.08% of the observations) Kappa= 0.365 SE of kappa = 0.017 95% confidence interval: From 0.332 to 0.398 The strength of agreement is considered to be 'fair'. Left Screening HFPTA to Left Diagnostic DPOAE A B Total A 2728 0427 3155 B 0528 0504 1032 Total 3256 931 4187 Number of observed agreements: 3232 ( 77.19% of the observations) Number of agreements expected by chance: 2682.9 ( 64.08% of the observations) Kappa= 0.365 SE of kappa = 0.017 95% confidence interval: From 0.332 to 0.398 The strength of agreement is considered to be 'fair'. Right Screening HFPTA to Right Diagnostic DPOAE A B Total A 2927 0529 3456 B 0327 0305 632 Total 3254 834 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2879.9 ( 70.45% of the observations) Kappa= 0.291 SE of kappa = 0.018 95% confidence interval: From 0.255 to 0.327 The strength of agreement is considered to be 'fair'. Right Screening DPOAE to Right Screening HFPTA A B Total A 2728 0427 3155 B 0528 0504 1032 Total 3256 931 4187 Number of observed agreements: 3232 ( 77.19% of the observations) Number of agreements expected by chance: 2682.9 ( 64.08% of the observations) Kappa= 0.365 SE of kappa = 0.017 95% confidence interval: From 0.332 to 0.398 The strength of agreement is considered to be 'fair'. Left Screening DPOAE to Left Screening HFPTA A B Total A 2927 0529 3456 B 0327 0305 632 Total 3254 834 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2879.9 ( 70.45% of the observations) Kappa= 0.291 SE of kappa = 0.018 95% confidence interval: From 0.255 to 0.327 The strength of agreement is considered to be 'fair'. Left Standard PTA to Left Screening DPOAE A B Total A 2927 0529 3456 B 0327 0305 632 Total 3254 834 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2879.9 ( 70.45% of the observations) Kappa= 0.291 SE of kappa = 0.018 95% confidence interval: From 0.255 to 0.327 The strength of agreement is considered to be 'fair'. Right Screening DPOAE to Right Standard PTA A B Total A 2728 0427 3155 B 0528 0504 1032 Total 3256 931 4187 Number of observed agreements: 3232 ( 77.19% of the observations) Number of agreements expected by chance: 2682.9 ( 64.08% of the observations) Kappa= 0.365 SE of kappa = 0.017 95% confidence interval: From 0.332 to 0.398 The strength of agreement is considered to be 'fair'. Left Screening DPOAE to Left Diagnostic HFPTA A B Total A 2727 0527 3254 B 0527 0505 1032 Total 3254 1032 4286 Number of observed agreements: 3232 ( 75.41% of the observations) Number of agreements expected by chance: 2719.0 ( 63.44% of the observations) Kappa= 0.327 SE of kappa = 0.016 95% confidence interval: From 0.295 to 0.360 The strength of agreement is considered to be 'fair'. Right Screening DPOAE to Right Diagnostic HFPTA A B Total A 2828 0428 3256 B 0428 0404 832 Total 3256 832 4088 Number of observed agreements: 3232 ( 79.06% of the observations) Number of agreements expected by chance: 2762.7 ( 67.58% of the observations) Kappa= 0.354 SE of kappa = 0.018 95% confidence interval: From 0.319 to 0.389 The strength of agreement is considered to be 'fair'. Left Screening DPOAE to Left Diagnostic DPOA A B Total A 2829 0328 3157 B 0429 0403 832 Total 3258 731 3989 Number of observed agreements: 3232 ( 81.02% of the observations) Number of agreements expected by chance: 2730.9 ( 68.46% of the observations) Kappa= 0.398 SE of kappa = 0.018 95% confidence interval: From 0.363 to 0.433 The strength of agreement is considered to be 'fair'. Right Screening DPOAE to Right Diagnostic DPOAE A B Total A 2524 0825 3349 B 0724 0708 1432 Total 3248 1533 4781 Number of observed agreements: 3232 ( 67.60% of the observations) Number of agreements expected by chance: 2734.3 ( 57.19% of the observations) Kappa= 0.243 SE of kappa = 0.015 95% confidence interval: From 0.214 to 0.272 The strength of agreement is considered to be 'fair'. Result :Overall Screening To Diagnostic Tests Right Standard PTA to Right Screening HFPTA Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 93.10 % 95% CI: 77.19 % to 98.95 % PositivePredictiveValue = 60.00 % (*) 95% CI: 15.40 % to 93.51 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Standard PTA to Left Screening HFPTA Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 93.10 % 95% CI: 77.19 % to 98.95 % PositivePredictiveValue = 60.00 % (*) 95% CI: 15.40 % to 93.51 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Right Diagnostic HFPTA to Right Screening HFPTA Sensitivity = 100.00 % 95% CI: 47.95 % to 100.00 % Specificity = 100.00 % 95% CI: 87.11 % to 100.00 % Positive Predictive Value = 100.00 % (*) 95% CI: 47.95 % to 100.00 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Diagnostic HFPTA to Left Screening HFPTA Sensitivity = 100.00 % 95% CI: 47.95 % to 100.00 % Specificity = 100.00 % 95% CI: 87.11 % to 100.00 % Positive Predictive Value = 100.00 % (*) 95% CI: 47.95 % to 100.00 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Diagnostic DPOAE to Left Screening HFPTA Sensitivity = 100.00 % 95% CI: 40.23 % to 100.00 % Specificity = 96.43 % 95% CI: 81.59 % to 99.40 % Positive Predictive Value = 80.00 % (*) 95% CI: 28.81 % to 96.70 % Negative PredictiveValue = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Right Diagnostic DPOAE to Right Screening HFPTA Sensitivity = 100.00 % 95% CI: 40.23 % to 100.00 % Specificity = 96.43 % 95% CI: 81.59 % to 99.40 % Positive Predictive Value = 80.00 % (*) 95% CI: 28.81 % to 96.70 % Negative PredictiveValue = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Right Screening DPOAE to Right Screening HFPTA Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 93.10 % 95% CI: 77.19 % to 98.95 % PositivePredictiveValue = 60.00 % (*) 95% CI: 15.40 % to 93.51 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Screening DPOAE to Left Screening HFPTA Sensitivity = 100.00 % 95% CI: 40.23 % to 100.00 % Specificity = 96.43 % 95% CI: 81.59 % to 99.40 % Positive = 80.00 % (*) 95% CI: 28.81 % to 96.70 % Negative PredictiveValue = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Standard PTA to Left Screening DPOAE Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 93.10 % 95% CI: 77.19 % to 98.95 % Positive PredictiveValue = 60.00 % (*) 95% CI: 15.40 % to 93.51 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Right Standard PTA to Right Screening DPOAE Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 93.10 % 95% CI: 77.19 % to 98.95 % PositivePredictiveValue = 60.00 % (*) 95% CI: 15.40 % to 93.51 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Diagnostic HFPTA to Left Screening DPOAE Sensitivity = 100.00 % 95% CI: 40.23 % to 100.00 % Specificity = 96.43 % 95% CI: 81.59 % to 99.40 % Positive Predictive Value = 80.00 % (*) 95% CI: 28.81 % to 96.70 % Negative PredictiveValue = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Right Diagnostic HFPTA to Right Screening DPOAE Sensitivity = 100.00 % 95% CI: 47.95 % to 100.00 % Specificity = 100.00 % 95% CI: 87.11 % to 100.00 % Positive Predictive Value = 100.00 % (*) 95% CI: 47.95 % to 100.00 % Negative Predictive Value = 100.00 % (*) 95% CI: 87.11 % to 100.00 % Left Diagnostic DPOA to Left Screening DPOAE Sensitivity = 100.00 % 95% CI: 40.23 % to 100.00 % Specificity = 100.00 % 95% CI: 87.54 % to 100.00 % Positive PredictiveValue = 100.00 % (*) 95% CI: 40.23 % to 100.00 % NegativePredictiveValue = 100.00 % (*) 95% CI: 87.54 % to 100.00 % Right Diagnostic DPOA to Right Screening DPOAE Sensitivity = 100.00 % 95% CI: 30.48 % to 100.00 % Specificity = 96.55 % 95% CI: 82.17 % to 99.42 % Positive PredictiveValue = 75.00 % (*) 95% CI: 20.34 % to 95.88 % Negative PredictiveValue = 100.00 % (*) 95% CI: 87.54 % to 100.00 % Result :Overall Screening To Diagnostic Tests Sensitivity = 100.00 % 95% CI: 58.93 % to 100.00 % Specificity = 96.00 % 95% CI: 79.58 % to 99.33 % Positive PredictiveValue = 87.50 % (*) 95% CI: 47.38 % to 97.93 % NegativePredictive= 100.00 % (*) 95% CI: 85.62 % to 100.00 % Discussion and Conclusion Simple and Multiple regression is a mode of multivariate statistics which predicts the most likely value(in this case probability of confirmed diagnosis which is a criterion or dependant variable) from the values of one(simple linear regression) and if two or more other variables called the predictors( in this case screening tests). It can only be predicted if the variables have significant correlation between each other ( correlation coefficient like Kendalls tau and Pearson ‘r’ should be statistically significant. Correlation Coefficient can range from a value of -1 to +1. Positive values means finding of diagnostic results are in line with screening results and vice versa. Types Of Regression Analysis A logistic regression is a statistical method for analyzing the dataset in which there are one or more independent variables that determine an outcome. The outcome is measured with a dichotomous variable (in which there are only two possible outcomes that is ectopic pregnancy or no ectopic pregnancy).Logistic regression thus can be used when the dependant variable is a nominal variable( variable that cannot me measured quantitatively)The goal of a logistic regression is to find out the best fitting (yet biologically reasonable) model to describe the relationship between the dichotomous characteristic of interest (dependent variable = response or outcome variable) and a set of independent (predictor or explanatory) variables. Logistic regression generates the coefficients (and its standard errors and significance levels) of a formula to predict a logit transformation of the probability of presence of a characteristic of interest: Y(Dependant Variable)= constant+ constantX(Independent Variable) This means whenever we assign a value of independent variable it gives the most likely value of a dependant variable. Logit(p)= a0 + b12.3X1 + b 13.2X2+ ……………..+bkXk where a0 is the Y intercept of the regression line and b12.3, b 13.2 and bk are partial regression coefficients where p is the probability of presence of the characteristic of interest. The logit transformation is calculated as the logged odds: Odds= p/1-p= probability of presence of an event/ probability of absence of an event And logit(p)= In[p/1-p] Ordinary Least squares (OLS) regression also predicts the chances or value of a criterion based on predictors but in this case the criterion or dependant variable should be an interval variable.(measurement variable which do not have a real 0 value) OLS Equation Y = ?0 + ?1X1 + ?2X2 + ?3D + e where the ?s are the OLS estimates of the Bs. OLS minimizes the sum of the squared residuals .OLS minimizes SUM e2.The residual, e, is the difference between the actual Y and the predicted Y and has a zero mean. In other words, OLS calculates the slope coefficients so that the difference between the predicted Y and the actual Y is minimized. (The residuals are squared in order to compare negative errors to positive errors more easily.) Cohen's kappa coefficient is a statistical measure of or inter-annotator agreement for qualitative (categorical) items. It is generally thought to be a more robust measure than simple percent agreement calculation since ? takes into account the agreement occurring by chance. Cohen's kappa measures the agreement between two raters who each classify N items into C mutually exclusive categories. The equation for ? is: where Pr(a) is the relative observed agreement among raters(in our case the diagnostic results) and Pr(e) is the hypothetical probability of chance agreement,(screening tests) using the observed data to calculate the probabilities of each observer randomly saying each category. If the raters are in complete agreement then ? = 1. If there is no agreement among results then ?=0.The regression equation has given the basis for the extent of relationship of one result with that of another. From the above principles that can be inferred that all the screening results highly tallies with the diagnostic findings as because wherever the comparisons had been made there is a high partial positive correlation co-efficients and they are significant , This is further confirmed by kappa analysis that the correlations are fair enough in all the occasions and hence there is a match between experimental probability distribution and theoretical probability distribution of findings and screenings.Thus results of screening audiology (hearing) tests to the results of diagnostic audiology tests indicate that screening tests are accurate enough like diagnostic tests in detecting ototoxicity ( reduction in hearing sensitivity that is caused by particular medications such as aminoglycosides antibiotics). Thus the screening HFPTA and screening DPOAE. audiology tests tested confirms the Results of diagnostic audiology tests which are the standard pure tone audiometry (PTA), diagnostic high frequency pure tone audiometry (HFPTA) and diagnostic distortion product otoacoustic emissions (DPOAE). The overall assessment of results was uniform in both male and females and when totality is concerned it reflects all the screening test results confirms diagnostic results of confirmed ototoxicity with aminoglycosides used for treatment of cystic fibrosis.However analysis of all results it is found that Left Standard PTA(Y) to Left Screening DPOAE(X) and Right Standard PTA to Right Screening DPOAE is the most correlated Diagnostic-Screening Test. There has been no errors in that assumption. The reliability of each analysis reflected that the kappa values at 95% CI is very close and thus confers reliability to our estimates. Confidence interval predicts the range of values that can be estimated in the population. Explanation of Sensitivity, Specificity,PPV and NPV (In our experimental context) Sensitivity: probability that a test result (screening) will be positive when the diagnosis (or another screening)is positive (true positive rate).= a / (a+b) Specificity: probability that a test result(screening) will be negative when the diagnosis (or another screening )does not confirm the same (true negative rate).= d / (c+d) Positive likelihood ratio: ratio between the probability of a positive test result given the presence of the diagnosis and the probability of a positive test result given the absence of the diagnosis, i.e.= True positive rate / False positive rate = Sensitivity / (1-Specificity) Negative likelihood ratio: ratio between the probability of a negative test result(screening) given the presence of the diagnosis(or another screening) and the probability of a negative test result given the absence of the diagnosis(or another screening, i.e.= False negative rate / True negative rate = (1-Sensitivity) / Specificity Positive predictive value: probability that the diagnosis is confirmed when the screening is positive. = a / (a+c) Negative predictive value: probability that the diagnosis is not confirmed when the screening test is negative. = d  / (b+d) If the sample sizes in the positive (Diagnosis present) and the negative (Diagnosis absent)) groups do not reflect the real prevalence of the disease, then the Positive and Negative predicted values cannot be estimated and you should ignore those values. Alternatively, when the diagnosis prevalence is known then the positive and negative predictive values can be calculated using the following formula's based on Bayes' theorem: PPV= sensitivity x prevalence/ sensitivity x prevalence + (1-specificity) x (1-prevalence) NPV= specificity x (1-prevalence)/ (1-sensitivity) x prevalence + specificity x (1-prevalence) In these analysis it reflects the specificity, likelihood ratios, PPV &NPV of all the matched tests are very high as far as screening and diagnosis are concerned but it does not reflect to tympanometric tests and otoscopic examinations.We can certainly interpret the screening tests gives us a option to predict ototoxicity where diagnosting testing becomes difficult. Reviewing the hypothesis that we wanted to test- Screening high frequency pure tone audiometry and distortion product otoacoustic emissions performed in outpatients’ clinics and inpatients wards using portable devices are highly sensitive, specific, and has high positive and negative predictive values which indicate enough to accurately detect early signs of aminoglycosides-induced hearing loss and those patients who are not ototoxic( rling out the chance) in comparison to comprehensive audiological tests performed in sound proof audiology clinic which is also confirmed by partial correlation coefficients as given by regression equations with relation to the actual diagnostic tests done, Hence these screening tests can be a good alternative to the diagnostic instruments which might not be feasible to be carried at test sites always. Moreover our analysis reflected that the screening tests were as effective in correlating to female and male patients (7/22 screened for ototoxicity versus 8/22 diagnostically confirmed for ototoxicity in females) and similarly (10/10 screened with ototoxicity exactly related with 10/10 patients who were not at all diagnosed with ototoxicity). From this study we can find the incidence of AG related ototoxicity in females(at least) as 34% (8/22 patients ototoxicity diagnosed) while males have a less incidence( in fact 0 in this study) of ototoxicity, Out of all the DPOAE screening and HFPTA was amongst the most sensitive tests with high reproducibility. References for Results Gardner IA, Greiner M (2006) Receiver-operating characteristic curves and likelihood ratios: improvements over traditional methods for the evaluation and application of veterinary clinical pathology tests. Veterinary Clinical Pathology, 35:8-17. Griner PF, Mayewski RJ, Mushlin AI, Greenland P (1981) Selection and interpretation of diagnostic tests and procedures. Annals of Internal Medicine, 94, 555-600. Hanley JA, McNeil BJ (1982) The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology, 143, 29-36. Diagnostic test calculator:http://www.medcalc.org/calc/diagnostic_test.php Metz CE (1978) Basic principles of ROC analysis. Seminars in Nuclear Medicine, 8, 283-298 Zhou XH, NA Obuchowski, DK McClish (2002) Statistical methods in diagnostic medicine. New York: Wiley. Zweig MH, Campbell G (1993) Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clinical Chemistry, 39, 561-577. Read More