eisenlohrmoullab / bata_nas

bata_nas

Please see the CLEAR Lab Scientific Manual to learn how to use the data files in the CLEAR Lab: https://docs.google.com/document/d/1D55irUON7pOKBVRUCCQbciXwiY5Cemi95ML85OAv6EM/edit#

Date of project start: 2022-02-22 Re-started work: 2023-09-26

Study Name: BATA

Project Name:BATA_NAS

Type of Project (Paper/Poster/Grant/Talk/DescribeOther): Paper

First author: Erin Walsh

Senior author: Tory Eisenlohr-Moul or Gabriel Dichter

Additional authors: Eisenlohr-Moul, Morrow, Dichter, Girdler?

Will this analytic project be preregistered (e.g., on OSF, clinicaltrials.gov)?: Yes, OSF

URL for preregistration (if applicable): https://osf.io/qha9e

Date of preregistration or prospectus approval by study PI: Feb 22, 2022

Slack Channel Link: https://clearlabcollab.slack.com/archives/G05UN40GMDE

bata_nas's Issues

examine whether SHAPS is importing correctly (enough observations?)

Look at IUD And CHC effects on steroid trajs

Recode lut vs all variable into between and within-person subvariables

Right now it conflates cycle phase (state) with being a naturally cycling AFAB (trait)

Plot new ratios across study to see outliers

Get info re immune exclusions for Erin

go get week 4 shaps from other datasets

Replace not detectable with limit of detection

prep trait dataset for descriptives

z-score BMI, age and merge into trait dataset, and also merge a full wide dataset of traits.

troubleshoot convergence issues in covar models (H1)

I'm seeing quite a few NPD and "did not converge" messages in the initial run. Consider evidence for random slopes.

Test interactions instead of ratios

write code for MLMs to test hypotheses

Redo temporal plots with reversed NAS ratios to see change over time

Plug NAS slopes into MLMs as moderators of self report trajectories (all obs)

H2 - calculate T4-T1 nas changes

Create results summary tables

create list of variables to keep based on preregistration, and make small/tidy dataset

Reconsider steroid trajectories given only pre-post timepoints

Add raw data over time plots and scatters

Add to macros

create H2 hypothesis test macro

NASprepostchange*time=outcome

create new reprostatus variable

create new variable that indicates luteal phase female vs. all others

increase units for ALLO so that random effects will run

multiply by a factor?

Consider creating aggregate NAS and inflammatory vars

create and plot descriptives for dataset

Replace ratios with interactions

Why only two obs per person in reg plots?

Plot steroids across luteal phase

Move dataset to brms, use template

review temporal plots and make decisions about outliers

see especially ALLO and THDOC - there are a few that are 100-1000x greater than all other obs.

troubleshoot 0 random intercept for SHAPS

re-merge SHAPS from Paul (see slack) to get missing observations!

rerun all with outliers autoremoved

Create simultaneous models w all predictors

ALLO + pregna outliers are due to luteal phases