Selected article for: "curve area and ROC analysis"

Title: 2017 ACVIM Forum Research Abstract Program
  • Document date: 2017_6_15
  • ID: ri2w5iby_384
    Snippet: The results indicate that none of the variables compared between the two groups were statistically significant (P-value < 0.05) in predicting the presence of disease. Although not significant, those variables close to a P-value of 0.10 all trend towards significance, and could represent possible predictive indices for disease: mean PLT and PCT were noticeably higher in IBD dogs and mean PCDW was noticeably lower in IBD dogs than in healthy contro.....
    Document: The results indicate that none of the variables compared between the two groups were statistically significant (P-value < 0.05) in predicting the presence of disease. Although not significant, those variables close to a P-value of 0.10 all trend towards significance, and could represent possible predictive indices for disease: mean PLT and PCT were noticeably higher in IBD dogs and mean PCDW was noticeably lower in IBD dogs than in healthy controls. The ROC analysis was performed to determine predictive value of each variable using area under the curve (AUC). Based on AUC values, no platelet indices are strongly correlated with disease. However, PLT, PCT, PCDW, MPM, and PMDW all have AUC values between 0.6 and 0.7, indicating they may be somewhat predictive of disease presence. Bivariate scatterplots with moving average demonstrate that only PLT (P-value 0.02) and PCT (P-value 0.03) were statistically significant in predicting disease severity. Linear regression showed mild correlation with CCECAI score for these two indices with PLT r 2 value of 0.28 and PCT of 0.25.

    Search related documents:
    Co phrase search for related documents