Merge - 'Ratio' Paired T-test

[dancoster]

Instead of Paired T-test calculate the 'Ratio' Paired T-Test:

1. For each 'before' and 'after', calculate it's ratio (after/before).
2. Than use standard one sample t-test with null hypotheiss of \mu =1.
3. Add Bonferoni.

[dancoster]

Exclude all pairs where (ratio==1) & (after_time<1), and than recalculate the p-values (all of them).

[PavanReddy28]

'Ratio' 1-sampled TTest for non-vital sign lab tests.

1. P-values - https://drive.google.com/file/d/1T1yIU8tF4sy3yT8Yeuj19TJ0gFhjt9pH/view?usp=sharing
2. Significant Pairs - https://drive.google.com/file/d/1TKHuWz69bW2q9DuJpKNzpifR-S27drEg/view?usp=sharing

[dancoster]

what do you mean by intersection? Did you exclusde - Exclude all pairs where (ratio==1) & (after_time<1), and than recalculate the p-values (all of them)?

[PavanReddy28]

Intersection of Bonferroni analysis, FDR analysis and hard threshold (alpha=0.01) analysis

[dancoster]

Add Bonferoni Correction and FDR

[PavanReddy28]

• Fixed the duplicate values issue.
• Added bonferroni and FDR corrected p values.

pvals_with_analysis_pairs_mimic_non_vital_signs.csv

sig_pairs_intersection_mimic_non_vital_signs.csv

[dancoster]

I think you didn't use Bonferroni and FDR on the 1-sample t-test? (That what's actually needed). For example:

Additionaly, there are many empty 'BonferroniPvals' and 'FDR Benjamini Corrected' values. Please send an adjusted file.

[dancoster]

Please also exclude inhuman values ('temp_mapping_140722_partial') before generating the p-value.

[PavanReddy28]

Update

• Fixed the bug related to missing bonferroni values.
• Pvals before inhuman values removal.

pvals_with_analysis_pairs_mimic_non_vital_signs.csv

sig_pairs_intersection_mimic_non_vital_signs.csv

[dancoster]

Looks good, thanks

[PavanReddy28]

Added ratio generation and pvalue calculation (using 1 Sampled Ttest code) as part of this commit