Associations of dementia polyexposure scores to Alzheimer's disease endophenotypes in a diverse population.

Code ↔ Paper

13 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 13 matches

[1] § METHODS › Clinical risk scores › Cognitive Health and Dementia Risk Reduction ↔ workflow/scripts/imputation.qmd, lines 55–114 · score 0.78 · atrial fibrillation, cognitive engagement, social support, insomnia, SD, obesity
[2] § METHODS › Clinical risk scores › Cognitive Health and Dementia Risk Reduction ↔ workflow/scripts/crs_analysis.qmd, lines 77–136 · score 0.77 · atrial fibrillation, cognitive engagement, social support, insomnia, obesity, stroke
[3] § METHODS › Statistical analysis › AD Endophenotypes ↔ workflow/scripts/crs_analysis.qmd, lines 1797–1943 · score 0.69 · cognitive function, plasma biomarkers, CogDrisk, ratio, hippocampal, executive
[4] § RESULTS › Higher clinical risk burden was associated with worsening AD endophenotypes › CogDRisk ↔ workflow/scripts/crs_analysis.qmd, lines 1758–1795 · score 0.68 · cortical thickness, hippocampal volume, verbal ability, executive function, CogDRisk
[5] § METHODS › Health and Aging Brain Study–Health Disparities › Endophenotype standardization ↔ workflow/scripts/crs_standardization.qmd, lines 199–252 · score 0.66 · composite scores, delayed, WMS, Animal, Digit, FAS
[6] § RESULTS › Higher clinical risk burden was associated with worsening AD endophenotypes › mCAIDE ↔ workflow/scripts/crs_analysis.qmd, lines 1758–1795 · score 0.58 · cortical thickness, hippocampal volumes, verbal ability, mCAIDE, race
[7] § METHODS › Health and Aging Brain Study–Health Disparities › Endophenotype standardization ↔ workflow/scripts/crs_standardization.qmd, lines 254–371 · score 0.57 · log transformed, raw, outliers, HABS, Plasma, NfL
[8] § METHODS › Statistical analysis › Prediction of pTau217/Aβ42 amyloid positivity by CRS ↔ workflow/scripts/cutoff_analysis.qmd, lines 25–44 · score 0.57 · optimal cutpoint, Youden, bootstrapping, pTau217, threshold
[9] § METHODS › Clinical risk scores › Lifestyle for Brain Health ↔ workflow/scripts/crs_analysis.qmd, lines 186–256 · score 0.56 · heart diseases, AUDIT, alcohol, renal, physical, RAPA
[10] § METHODS › Clinical risk scores › Lifestyle for Brain Health ↔ workflow/scripts/imputation.qmd, lines 164–233 · score 0.56 · heart diseases, AUDIT, alcohol, renal, physical, RAPA
[11] § METHODS › Statistical analysis › Cognitive impairment ↔ workflow/scripts/crs_analysis.qmd, lines 2227–2312 · score 0.54 · Logistic regression, CogDRisk, squared, Nagelkerke, mCAIDE, MCI
[12] § METHODS › Health and Aging Brain Study–Health Disparities › Endophenotype standardization ↔ workflow/scripts/crs_standardization.qmd, lines 376–438 · score 0.54 · Cortical thickness, entorhinal, fusiform, inferior, Hippocampal, AD
[13] § RESULTS › Higher clinical risk burden was associated with worsening AD endophenotypes › Lifestyle for Brain Health ↔ workflow/scripts/crs_analysis.qmd, lines 1797–1943 · score 0.53 · plasma biomarkers, hyperintensity, population, hippocampal, executive, neuroimaging

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The paper is loaded by your browser, with JavaScript: read it at doi.org or on Europe PMC.

The authors' code

Quarto · 2,407 lines · 113 KB · MIT · 7 matches

---
title: "CRS analysis"
author: "Meri Okorie"
date: "2025-12-10"
output: html_document
---
### Setup
```{r}
library(ggplot2)
library(dplyr)
library(tidyr)
library(pROC)
library(PRROC)
library(readr)
library(Metrics)
library(r2redux)
library(ggalluvial)
library(glue)
library(broom)
library(purrr)
library(tidyverse)
library(scales)
library(rcompanion)
library(plotgardener)
library(cowplot)
library(parameters)
library(mice)
library(missForest)
library(rcompanion)
library(forcats)
```
### Imputation
```{r}
#former smoker
data_org$smoke_former = ifelse(data_org$substance_use_smoke_end_age < data_org$age, 1, 0)
data_org$smoke_former[is.na(data_org$smoke_former)] = 0
new_data_cols = c("age", "sex", "id_education", "id_income", "id_language_primary", "bmi", "imh_diabetes", "imh_stroke", "imh_tbi",
"cdx_hypertension", "cdx_depression", "rapa_1_1", "rapa_1_2", "rapa_1_6", "rapa_1_7", "smoke_ever", "smoke_former", "smoke_currently",
"social_support_total", "audit_1", "eGFR", "imh_high_cholesterol", "bw_hdl_chol", "bw_chol_total", "om_bp1_sys", "om_bp2_sys",
"rapa_1_total", "rapa_2_total", "race", "cdr", "hc_insurance_no_insurance", "health_status", "prs_z_prscsx", "PC1", "PC2", "PC3", "PC4")
vars = c("bmi", "imh_diabetes", "imh_tbi", "rapa_1_1", "rapa_1_2", "rapa_1_6", "rapa_1_total", "rapa_2_total",
"rapa_1_7", "smoke_ever", "smoke_currently", "social_support_total", "audit_1",
"eGFR", "bw_hdl_chol", "om_bp1_sys", "om_bp2_sys", "bw_chol_total")
data_for_imp_fixed = data_org %>%
select(all_of(new_data_cols)) %>%
mutate(across(where(is.character), as.factor)) %>%
mutate(across(where(~ is.numeric(.) && n_distinct(., na.rm = TRUE) == 2), ~ as.factor(.))) %>%
as.data.frame()
set.seed(123)
imp_mf = missForest(data_for_imp_fixed, maxiter = 10, ntree = 100)
data_imp = imp_mf$ximp
data_imp_values = imp_mf$ximp
data_imputed = data_org
data_imputed[vars] = data_imp_values[vars]
plot(density(data_org$rapa_2_total, na.rm = T), xlab = "BMI", lwd = 2, col = "black")
lines(density(data_imputed$rapa_2_total), col = "red", lty = 3, lwd = 2)
# CogD
threshold_ss = quantile(data_imputed$social_support_total, 0.15, na.rm = TRUE)
habs_imp = data_imputed %>%
mutate(
bw_chol_total_mmoll = bw_chol_total / 38.67
)
data.cogd_imp = habs_imp %>% # insomnia, fish consumption, atrial fibrillation, and cognitive engagement not measured
mutate(
cogd_age = case_when(
sex == 0 & age < 60 ~ 0,
sex == 0 & age >= 60 & age <= 64 ~ 0,
sex == 0 & age >= 65 & age <= 69 ~ 5,
sex == 0 & age >= 70 & age <= 74 ~ 8,
sex == 0 & age >= 75 & age <= 79 ~ 12,
sex == 0 & age >= 80 & age <= 84 ~ 17,
sex == 0 & age >= 85 & age <= 89 ~ 20,
sex == 0 & age >= 90 ~ 22,
sex == 1 & age < 60 ~ 0,
sex == 1 & age >= 60 & age <= 64 ~ 0,
sex == 1 & age >= 65 & age <= 69 ~ 5,
sex == 1 & age >= 70 & age <= 74 ~ 7,
sex == 1 & age >= 75 & age <= 79 ~ 13,
sex == 1 & age >= 80 & age <= 84 ~ 16,
sex == 1 & age >= 85 & age <= 89 ~ 19,
sex == 1 & age >= 90 ~ 23,
TRUE ~ NA_real_
),
cogd_edu = case_when(
id_education > 11 ~ 0,
id_education >= 8 & id_education <= 11 ~ 2,
id_education < 8 ~ 4,
TRUE ~ NA_real_
),
cogd_obesity = case_when( # only assigned for patients under the age of 65
bmi >= 18.5 & bmi <= 25 ~ 0, # normal
bmi > 25 & bmi <= 30 ~ 1, # overweight
bmi < 18.5 ~ 3, # underweight
bmi > 30 ~ 2, # obese
TRUE ~ NA_real_
),
cogd_chol = case_when(
bw_chol_total_mmoll < 6.5 ~ 0,
bw_chol_total_mmoll >= 6.5 ~ 3,
TRUE ~ NA_real_
),
cogd_diabetes = case_when(
imh_diabetes == 0 ~ 0,
sex == 0 & imh_diabetes == 1 ~ 2,
sex == 1 & imh_diabetes == 1 ~ 3,
TRUE ~ NA_real_
),
cogd_stroke = case_when(
imh_stroke == 0 ~ 0,
imh_stroke == 1 ~ 2,
TRUE ~ NA_real_
),
cogd_tbi = case_when(
imh_tbi == 0 ~ 0,
imh_tbi == 1 ~ 1,
TRUE ~ NA_real_
),
cogd_hypertension = case_when(
cdx_hypertension == 0 ~ 0,
cdx_hypertension == 1 ~ 1,
TRUE ~ NA_real_
),
cogd_depression = case_when(
cdx_depression == 0 ~ 0,
cdx_depression == 1 ~ 4,
TRUE ~ NA_real_
),
cogd_physical = case_when(
rapa_1_6 == 1 | rapa_1_7 == 1 ~ -3,
is.na(rapa_1_6) & is.na(rapa_1_7) ~ NA_real_,
TRUE ~ 0
),
cogd_smoke = case_when(
smoke_ever == 0 ~ 0,
smoke_former == 0 ~ 0,
smoke_former == 1 ~ 0.2,
smoke_currently == 0 ~ 0,
smoke_currently == 1 ~ 2,
TRUE ~ NA_real_
),
cogd_social = case_when(
social_support_total >= threshold_ss ~ 2,
social_support_total < threshold_ss ~ 0,
TRUE ~ NA_real_
)
) %>%
rowwise() %>%
mutate(
# Compute cogd score (else condition) and set it to NA if any variable is NA (if condition)
cogd = if(any(is.na(c(cogd_age, cogd_edu, cogd_obesity, cogd_chol, cogd_diabetes, cogd_stroke, cogd_tbi,
cogd_hypertension, cogd_depression, cogd_physical, cogd_smoke, cogd_social)))) {
NA_real_
} else {
sum(c(cogd_age, cogd_edu, cogd_obesity, cogd_chol, cogd_diabetes, cogd_stroke, cogd_tbi,
cogd_hypertension, cogd_depression, cogd_physical, cogd_smoke, cogd_social), na.rm = TRUE)
}
) %>%
ungroup()
# mutate(cogd_noedu = ifelse(!is.na(cogd) & !is.na(cogd_edu), cogd - cogd_edu, NA_real_))
# Libra
data.libra_imp = data.cogd_imp %>%
mutate(
libra_obesity = case_when( # only assigned for patients under the age of 65
bmi < 18.5 ~ 0,
bmi >= 18.5 & bmi < 25 ~ 0, # normal
bmi >= 25 & bmi < 30 ~ 1.6, # overweight
bmi >= 30 ~ 1.6, # obese
TRUE ~ NA_real_
),
libra_chol = case_when(
bw_chol_total_mmoll < 6.5 ~ 0,
bw_chol_total_mmoll >= 6.5 ~ 1.4,
TRUE ~ NA_real_
),
libra_diabetes = case_when(
imh_diabetes == 0 ~ 0,
imh_diabetes == 1 ~ 1.3,
TRUE ~ NA_real_
),
libra_hypertension = case_when(
cdx_hypertension == 0 ~ 0,
cdx_hypertension == 1 ~ 1.6,
TRUE ~ NA_real_
),
libra_depression = case_when(
cdx_depression == 0 ~ 0,
cdx_depression == 1 ~ 2.1,
TRUE ~ NA_real_
),
libra_physical = case_when( # 0.27% missigness
rapa_1_1 == 1 | rapa_1_2 == 1 ~ 1.1,
is.na(rapa_1_1) & is.na(rapa_1_2) ~ NA_real_,
TRUE ~ 0
),
libra_smoke = case_when(
smoke_currently == 1 ~ 1.5,
smoke_currently == 0 ~ 0,
TRUE ~ NA_real_
),
libra_alcohol = case_when(
audit_1 == 3 ~ -1.0,
!is.na(audit_1) ~ 0, # for all other non-missing values
TRUE ~ NA_real_
),
libra_renal = case_when(
eGFR < 60 ~ 1.1,
eGFR >= 60 ~ 0,
TRUE ~ NA_real_
),
libra_heart = case_when(
cdp_heart_disease == 1 | imh_stroke == 1 | cdp_myocardial == 1 ~ 1.0,
is.na(cdp_heart_disease) & is.na(imh_stroke) & is.na(cdp_myocardial) ~ NA_real_,
TRUE ~ 0.0
)) %>%
rowwise() %>%
mutate(
libra = if (any(is.na(c_across(c(
libra_obesity, libra_chol, libra_diabetes, libra_hypertension, libra_depression,
libra_physical, libra_smoke, libra_alcohol, libra_renal, libra_heart
))))) {
NA_real_
} else {
sum(c_across(c(
libra_obesity, libra_chol, libra_diabetes, libra_hypertension, libra_depression,
libra_physical, libra_smoke, libra_alcohol, libra_renal, libra_heart
)))
}
) %>%
ungroup()
#mCaide
data.libra_imp = data.libra_imp %>%
mutate(rapa_total = rapa_1_total + rapa_2_total)
threshold_rapa = quantile(data.libra_imp$rapa_total, probs = 1/3, na.rm = TRUE)
data.libra_imp$om_bp_sys_avg = (data.libra_imp$om_bp1_sys + data.libra_imp$om_bp2_sys) / 2
data.caide_imp = data.libra_imp %>%
mutate(
mcaide_age = case_when(
age <= 64 ~ 0,
age >= 65 & age < 73 ~ 1,
age >= 73 ~ 2,
TRUE ~ NA_real_
),
mcaide_educ = case_when(
id_education < 12 ~ 2,
id_education >= 12 & id_education <= 16 ~ 1,
id_education > 16 ~ 0,
TRUE ~ NA_real_
),
mcaide_sex = case_when(
sex == 0 ~ 1,
sex == 1 ~ 0,
TRUE ~ NA_real_
),
mcaide_bmi = case_when(
bmi <= 30 ~ 0,
bmi > 30 ~ 2,
TRUE ~ NA_real_
),
mcaide_sbp = case_when(
om_bp_sys_avg < 140 ~ 0,
om_bp_sys_avg >= 140 ~ 2,
TRUE ~ NA_real_
),
mcaide_chol = case_when(
imh_high_cholesterol == 0 ~ 0,
imh_high_cholesterol == 1 ~ 2,
TRUE ~ NA_real_
),
mcaide_phy = case_when(
rapa_total < threshold_rapa ~ 2,
rapa_total >= threshold_rapa ~ 0,
TRUE ~ NA_real_
),
) %>%
rowwise() %>%
mutate(
mcaide = if(any(is.na(c(mcaide_age, mcaide_educ, mcaide_sex, mcaide_bmi, mcaide_sbp, mcaide_chol, mcaide_phy)))) {
NA_real_
} else {
sum(c(mcaide_age, mcaide_educ, mcaide_sex, mcaide_bmi, mcaide_sbp, mcaide_chol, mcaide_phy), na.rm = TRUE)
}
) %>%
ungroup()
# WHICAP
data_all_imp = data.caide_imp %>%
mutate(
whicap_sex = case_when(
sex == 0 ~ 0,
sex == 1 ~ 1,
TRUE ~ NA_real_
),
whicap_age = case_when(
age <= 70 ~ 0,
age > 70 & age <= 75 ~ 6,
age > 75 & age <= 80 ~ 8,
age > 80 & age <= 85 ~ 13,
age > 85 ~ 21,
TRUE ~ NA_real_
),
whichap_diabetes = case_when(
cdx_diabetes == 0 ~ 0,
cdx_diabetes ==1 ~ 3,
TRUE ~ NA_real_
),
whicap_hypertension = case_when(
cdx_hypertension == 0 ~ 0,
cdx_hypertension == 1 ~ 1,
TRUE ~ NA_real_
),
whicap_smoker = case_when(
smoke_currently == 0 ~ 0,
smoke_currently == 1 ~ 5,
TRUE ~ NA_real_
),
whicap_cholesterol = case_when(
sex == 1 & bw_hdl_chol < 50 ~ 3,
sex == 1 & bw_hdl_chol >= 50 ~ 0,
sex == 0 & bw_hdl_chol < 40 ~ 3,
sex == 0 & bw_hdl_chol >= 40 ~ 0,
TRUE ~ NA_real_
),
whicap_bmi = case_when(
bmi > 25 ~ 7,
bmi <= 25 ~ 0,
TRUE ~ NA_real_
),
whicap_edu = case_when(
id_education > 9 ~ 0,
id_education >= 7 & id_education <= 9 ~ 8,
id_education <= 6 ~ 11,
TRUE ~ NA_real_
),
whicap_race = case_when(
race == "NHW" ~ 0,
race == "Hispanic" ~ 4,
race == "Black" ~ 5,
TRUE ~ NA_real_
)) %>%
rowwise() %>%
mutate(
whicap = if(any(is.na(c(whicap_sex, whicap_age, whichap_diabetes, whicap_hypertension, whicap_smoker,
whicap_cholesterol, whicap_bmi, whicap_edu, whicap_race)))) {
NA_real_
} else {
sum(c(whicap_sex, whicap_age, whichap_diabetes, whicap_hypertension, whicap_smoker,
whicap_cholesterol, whicap_bmi, whicap_edu, whicap_race), na.rm = TRUE)
}
) %>%
ungroup()
```
```{r}
dat = data_all_imp %>%
as.data.frame() %>%
mutate(
cdx_cn_vs_ci = case_when(
cdx_cog == 0 ~ 1,
cdx_cog %in% c(1, 2) ~ 0
),
cdx_mci_vs_cn = case_when(
cdx_cog == 0 ~ 0,
cdx_cog == 1 ~ 1,
TRUE ~ NA # drop dementia
),
cdx_dem_vs_cn = case_when(
cdx_cog == 0 ~ 0,
cdx_cog == 2 ~ 1,
TRUE ~ NA # drop MCI
),
cdx_ci_vs_cn = case_when(
cdx_cog == 0 ~ 0,
cdx_cog %in% c(1, 2) ~ 1, # MCI + dementia
TRUE ~ NA
)
)
data = dat %>%
rowwise() %>%
mutate(
cogd_sva = if (any(is.na(c_across(c(
cogd_obesity, cogd_chol, cogd_diabetes, cogd_stroke, cogd_tbi, cogd_hypertension, cogd_depression, cogd_physical, cogd_smoke, cogd_social
))))) {
NA_real_
} else {
sum(c_across(c(
cogd_obesity, cogd_chol, cogd_diabetes, cogd_stroke, cogd_tbi, cogd_hypertension, cogd_depression, cogd_physical, cogd_smoke, cogd_social
)))
},
mcaide_sva = if (any(is.na(c_across(c(mcaide_bmi, mcaide_sbp, mcaide_chol, mcaide_phy))))) {
NA_real_
} else {
sum(c_across(c(mcaide_bmi, mcaide_sbp, mcaide_chol, mcaide_phy)))
},
whicap_sva = if (any(is.na(c_across(c(
whichap_diabetes, whicap_hypertension, whicap_smoker, whicap_cholesterol, whicap_bmi, whicap_race
))))) {
NA_real_
} else {
sum(c_across(c(
whichap_diabetes, whicap_hypertension, whicap_smoker, whicap_cholesterol, whicap_bmi, whicap_race
)))
}
) %>%
mutate(
libra_age_sva = case_when(
sex == 0 & age < 65 ~ 0,
sex == 0 & age >= 65 & age < 69 ~ 0.13,
sex == 0 & age >= 70 & age <= 74 ~ 1.57,
sex == 0 & age >= 75 & age <= 79 ~ 2.04,
sex == 0 & age >= 80 & age <= 84 ~ 3.37,
sex == 0 & age >= 85 & age <= 89 ~ 4.24,
sex == 0 & age >= 90 ~ 4.93,
sex == 1 & age < 65 ~ 0,
sex == 1 & age >= 65 & age <= 69 ~ 0.64,
sex == 1 & age >= 70 & age <= 74 ~ 1.87,
sex == 1 & age >= 75 & age <= 79 ~ 2.75,
sex == 1 & age >= 80 & age <= 84 ~ 3.71,
sex == 1 & age >= 85 & age <= 89 ~ 4.58,
sex == 1 & age >= 90 ~ 5.28,
TRUE ~ NA_real_
),
libra_edu_sva = case_when(
id_education > 11 ~ 0,
id_education >= 8 & id_education >= 11 ~ 0.42,
id_education < 8 ~ 0.80,
TRUE ~ NA_real_
)) %>%
rowwise() %>%
mutate(
libra_sva = if (any(is.na(c_across(c(
libra_age_sva, libra_edu_sva, libra_obesity, libra_chol, libra_diabetes, libra_hypertension,
libra_depression, libra_physical, libra_smoke, libra_alcohol, libra_renal, libra_heart
))))) {
NA_real_
} else {
sum(c_across(c(
libra_age_sva, libra_edu_sva, libra_obesity, libra_chol, libra_diabetes, libra_hypertension,
libra_depression, libra_physical, libra_smoke, libra_alcohol, libra_renal, libra_heart
)))
}
) %>%
ungroup() %>%
rename_with(~ paste0(.x, "_org"),
.cols = c(cogd, mcaide, libra, whicap,
cogd_sva, mcaide_sva, libra_sva, whicap_sva)) %>%
mutate(across(ends_with("_org"),
~ as.numeric(scale(.)),
.names = "{sub('_org$', '', .col)}"))
```
## Regression
### Linear - All
```{r}
all_num_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'ef', 'all', 'base', 'interview_language + age + sex + id_education + apoe',
'em', 'all', 'base', 'interview_language + age + sex + id_education + apoe',
'va', 'all', 'base', 'interview_language + age + sex + id_education + apoe',
'mmse_total', 'all', 'base', 'apoe + age + sex + id_education + interview_language',
'cdr', 'all', 'base', 'apoe + age + sex + id_education + interview_language',
'ab40', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42_ab40', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'tau', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'ptau', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'nfl', 'all', 'base', 'apoe + age + sex + bmi + eGFR',
'meta_roi', 'all', 'base', 'apoe + age + sex + icv',
'z_hippcampul_vol', 'all', 'base', 'apoe + age + sex + icv',
'wmh_volume_log', 'all', 'base', 'apoe + age + sex + icv',
#Demographics - apoe
'ef', 'all', 'base_noapoe', 'interview_language + age + sex + id_education',
'em', 'all', 'base_noapoe', 'interview_language + age + sex + id_education',
'va', 'all', 'base_noapoe', 'interview_language + age + sex + id_education',
'mmse_total', 'all', 'base_noapoe', 'age + sex + id_education + interview_language',
'cdr', 'all', 'base_noapoe', 'age + sex + id_education + interview_language',
'ab40', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'ab42', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'ab42_ab40', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'tau', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'ptau', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'nfl', 'all', 'base_noapoe', 'age + sex + bmi + eGFR',
'meta_roi', 'all', 'base_noapoe', 'age + sex + icv',
'z_hippcampul_vol', 'all', 'base_noapoe', 'age + sex + icv',
'wmh_volume_log', 'all', 'base_noapoe', 'age + sex + icv',
#mCAIDE
'ef', 'all', 'mcaide', 'interview_language + apoe + mcaide',
'em', 'all', 'mcaide', 'interview_language + apoe + mcaide',
'va', 'all', 'mcaide', 'interview_language + apoe + mcaide',
'mmse_total', 'all', 'mcaide', 'interview_language + apoe + mcaide',
'cdr', 'all', 'mcaide', 'interview_language + apoe + mcaide',
'ab40', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'ab42', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'ab42_ab40', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'tau', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'ptau', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'nfl', 'all', 'mcaide', 'apoe + eGFR + mcaide',
'meta_roi', 'all', 'mcaide', 'apoe + icv + mcaide',
'z_hippcampul_vol', 'all', 'mcaide', 'apoe + icv + mcaide',
'wmh_volume_log', 'all', 'mcaide', 'apoe + icv + mcaide',
# LIBRA
'ef', 'all', 'libra', 'interview_language + apoe + libra',
'em', 'all', 'libra', 'interview_language + apoe + libra',
'va', 'all', 'libra', 'interview_language + apoe + libra',
'mmse_total', 'all', 'libra', 'interview_language + apoe + libra',
'cdr', 'all', 'libra', 'interview_language + apoe + libra',
'ab40', 'all', 'libra', 'apoe + libra',
'ab42', 'all', 'libra', 'apoe + libra',
'ab42_ab40', 'all', 'libra', 'apoe + libra',
'tau', 'all', 'libra', 'apoe + libra',
'ptau', 'all', 'libra', 'apoe + libra',
'nfl', 'all', 'libra', 'apoe + libra',
'meta_roi', 'all', 'libra', 'apoe + icv + libra',
'z_hippcampul_vol', 'all', 'libra', 'apoe + icv + libra',
'wmh_volume_log', 'all', 'libra', 'apoe + icv + libra',
# WHICAP
'ef', 'all', 'whicap', 'interview_language + apoe + whicap',
'em', 'all', 'whicap', 'interview_language + apoe + whicap',
'va', 'all', 'whicap', 'interview_language + apoe + whicap',
'mmse_total', 'all', 'whicap', 'interview_language + apoe + whicap',
'cdr', 'all', 'whicap', 'interview_language + apoe + whicap',
'ab40', 'all', 'whicap', 'apoe + eGFR + whicap',
'ab42', 'all', 'whicap', 'apoe + eGFR + whicap',
'ab42_ab40', 'all', 'whicap', 'apoe + eGFR + whicap',
'tau', 'all', 'whicap', 'apoe + eGFR + whicap',
'ptau', 'all', 'whicap', 'apoe + eGFR + whicap',
'nfl', 'all', 'whicap', 'apoe + eGFR + whicap',
'meta_roi', 'all', 'whicap', 'apoe + icv + whicap',
'z_hippcampul_vol', 'all', 'whicap', 'apoe + icv + whicap',
'wmh_volume_log', 'all', 'whicap', 'apoe + icv + whicap',
# CogD
'ef', 'all', 'cogd', 'interview_language + apoe + cogd',
'em', 'all', 'cogd', 'interview_language + apoe + cogd',
'va', 'all', 'cogd', 'interview_language + apoe + cogd',
'mmse_total', 'all', 'cogd', 'interview_language + apoe + cogd',
'cdr', 'all', 'cogd', 'interview_language + apoe + cogd',
'ab40', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42_ab40', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'tau', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'ptau', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'nfl', 'all', 'cogd', 'apoe + bmi + eGFR + cogd',
'meta_roi', 'all', 'cogd', 'apoe + icv + cogd',
'z_hippcampul_vol', 'all', 'cogd', 'apoe + icv + cogd',
'wmh_volume_log', 'all', 'cogd', 'apoe + icv + cogd',
# Sensitivity analysis
#LIBRA
'ef', 'all', 'libra_sva', 'interview_language + apoe + libra_sva',
'em', 'all', 'libra_sva', 'interview_language + apoe + libra_sva',
'va', 'all', 'libra_sva', 'interview_language + apoe + libra_sva',
'mmse_total', 'all', 'libra_sva', 'interview_language + apoe + libra_sva',
'cdr', 'all', 'libra_sva', 'interview_language + apoe + libra_sva',
'ab40', 'all', 'libra_sva', 'apoe + libra_sva',
'ab42', 'all', 'libra_sva', 'apoe + libra_sva',
'ab42_ab40', 'all', 'libra_sva', 'apoe + libra_sva',
'tau', 'all', 'libra_sva', 'apoe + libra_sva',
'ptau', 'all', 'libra_sva', 'apoe + libra_sva',
'nfl', 'all', 'libra_sva', 'apoe + libra_sva',
'meta_roi', 'all', 'libra_sva', 'apoe + icv + libra_sva',
'z_hippcampul_vol', 'all', 'libra_sva', 'apoe + icv + libra_sva',
'wmh_volume_log', 'all', 'libra_sva', 'apoe + icv + libra_sva',
#mCAIDE
'ef', 'all', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'em', 'all', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'va', 'all', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'mmse_total', 'all', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'cdr', 'all', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'ab40', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42_ab40', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'tau', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ptau', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'nfl', 'all', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'meta_roi', 'all', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'z_hippcampul_vol', 'all', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'wmh_volume_log', 'all', 'mcaide_sva', 'apoe + icv + mcaide_sva',
#WHICAP
'ef', 'all', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'em', 'all', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'va', 'all', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'mmse_total', 'all', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'cdr', 'all', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'ab40', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42_ab40', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'tau', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ptau', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'nfl', 'all', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'meta_roi', 'all', 'whicap_sva', 'apoe + icv + whicap_sva',
'z_hippcampul_vol', 'all', 'whicap_sva', 'apoe + icv + whicap_sva',
'wmh_volume_log', 'all', 'whicap_sva', 'apoe + icv + whicap_sva',
#CogD
'ef', 'all', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'em', 'all', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'va', 'all', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'mmse_total', 'all', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'cdr', 'all', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'ab40', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42_ab40', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'tau', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ptau', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'nfl', 'all', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'meta_roi', 'all', 'cogd_sva', 'apoe + icv + cogd_sva',
'z_hippcampul_vol', 'all', 'cogd_sva', 'apoe + icv + cogd_sva',
'wmh_volume_log', 'all', 'cogd_sva', 'apoe + icv + cogd_sva'
) %>%
mutate(eq = glue('{outcome} ~ {predictors}'),
res = map(eq, lm, data = data),
dataf = map(res, tidy),
mod = map(res, glance),
r2 = map_dbl(mod, ~ .x$r.squared),
r2.adj = map_dbl(mod, ~ .x$adj.r.squared),
n = map_int(res, nobs))
```
### CDX - All
```{r}
all_cdx_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'cdx_cn_vs_ci', 'all', 'base', 'age + sex + apoe + id_education',
'cdx_mci_vs_cn', 'all', 'base', 'age + sex + apoe + id_education',
'cdx_dem_vs_cn', 'all', 'base', 'age + sex + apoe + id_education',
'cdx_ci_vs_cn', 'all', 'base', 'age + sex + apoe + id_education',
#Demographics
'cdx_cn_vs_ci', 'all', 'base_noapoe', 'age + sex + id_education',
'cdx_mci_vs_cn', 'all', 'base_noapoe', 'age + sex + id_education',
'cdx_dem_vs_cn', 'all', 'base_noapoe', 'age + sex + id_education',
'cdx_ci_vs_cn', 'all', 'base_noapoe', 'age + sex + id_education',
#LIBRA
'cdx_cn_vs_ci', 'all', 'libra', 'apoe + libra',
'cdx_mci_vs_cn', 'all', 'libra', 'apoe + libra',
'cdx_dem_vs_cn', 'all', 'libra', 'apoe + libra',
'cdx_ci_vs_cn', 'all', 'libra', 'apoe + libra',
#mCAIDE
'cdx_cn_vs_ci', 'all', 'mcaide', 'apoe + mcaide',
'cdx_mci_vs_cn', 'all', 'mcaide', 'apoe + mcaide',
'cdx_dem_vs_cn', 'all', 'mcaide', 'apoe + mcaide',
'cdx_ci_vs_cn', 'all', 'mcaide', 'apoe + mcaide',
#WHICAP
'cdx_cn_vs_ci', 'all', 'whicap', 'apoe + whicap',
'cdx_mci_vs_cn', 'all', 'whicap', 'apoe + whicap',
'cdx_dem_vs_cn', 'all', 'whicap', 'apoe + whicap',
'cdx_ci_vs_cn', 'all', 'whicap', 'apoe + whicap',
#CogD
'cdx_cn_vs_ci', 'all', 'cogd', 'apoe + cogd',
'cdx_mci_vs_cn', 'all', 'cogd', 'apoe + cogd',
'cdx_dem_vs_cn', 'all', 'cogd', 'apoe + cogd',
'cdx_ci_vs_cn', 'all', 'cogd', 'apoe + cogd',
# Sensitivity analysis
#LIBRA
'cdx_cn_vs_ci', 'all', 'libra_sva', 'apoe + libra_sva',
'cdx_mci_vs_cn', 'all', 'libra_sva', 'apoe + libra_sva',
'cdx_dem_vs_cn', 'all', 'libra_sva', 'apoe + libra_sva',
'cdx_ci_vs_cn', 'all', 'libra_sva', 'apoe + libra_sva',
#mCAIDE
'cdx_cn_vs_ci', 'all', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_mci_vs_cn', 'all', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_dem_vs_cn', 'all', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_ci_vs_cn', 'all', 'mcaide_sva', 'apoe + mcaide_sva',
#WHICAP
'cdx_cn_vs_ci', 'all', 'whicap_sva', 'apoe + whicap_sva',
'cdx_mci_vs_cn', 'all', 'whicap_sva', 'apoe + whicap_sva',
'cdx_dem_vs_cn', 'all', 'whicap_sva', 'apoe + whicap_sva',
'cdx_ci_vs_cn', 'all', 'whicap_sva', 'apoe + whicap_sva',
#CogD
'cdx_cn_vs_ci', 'all', 'cogd_sva', 'apoe + cogd_sva',
'cdx_mci_vs_cn', 'all', 'cogd_sva', 'apoe + cogd_sva',
'cdx_dem_vs_cn', 'all', 'cogd_sva', 'apoe + cogd_sva',
'cdx_ci_vs_cn', 'all', 'cogd_sva', 'apoe + cogd_sva'
) %>%
mutate(
eq = glue("{outcome} ~ {predictors}"),
res = map(eq, ~ glm(.x, data = data, family = "binomial")),
auc = map_dbl(res, ~ {
y = model.response(model.frame(.x)) # create model.frame(x) which is the exact same data as res
p = predict(.x, type = "response") # predict fitted probabilities for each outcome
if (length(unique(y)) > 1) { # if outcome is all 0s or 1s, skip
as.numeric(pROC::auc(pROC::roc(y, p))) # compute ROC curve then AUC and return numeric number (as.numeric)
} else {NA_real_}}), # if the outcome has only one class, return NA
nagelkerke_r2 = map_dbl(res, ~ {
out = tryCatch(performance::r2_nagelkerke(.x), error = function(e) NA_real_)
as.numeric(out)}),
dataf = map(res, tidy),
mod = map(res, glance),
n = map_int(res, nobs))
```
### Linear - Ancestry
```{r}
eur_num_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'ef', 'NHW', 'base', 'interview_language + age + sex + id_education + apoe',
'em', 'NHW', 'base', 'interview_language + age + sex + id_education + apoe',
'va', 'NHW', 'base', 'interview_language + age + sex + id_education + apoe',
'mmse_total', 'NHW', 'base', 'apoe + age + sex + id_education + interview_language',
'cdr', 'NHW', 'base', 'apoe + age + sex + id_education + interview_language',
'ab40', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42_ab40', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'tau', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'ptau', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'nfl', 'NHW', 'base', 'apoe + age + sex + bmi + eGFR',
'meta_roi', 'NHW', 'base', 'apoe + age + sex + icv',
'z_hippcampul_vol', 'NHW', 'base', 'apoe + age + sex + icv',
'wmh_volume_log', 'NHW', 'base', 'apoe + age + sex + icv',
#Demographics - apoe
'ef', 'NHW', 'base_noapoe', 'interview_language + age + sex + id_education',
'em', 'NHW', 'base_noapoe', 'interview_language + age + sex + id_education',
'va', 'NHW', 'base_noapoe', 'interview_language + age + sex + id_education',
'mmse_total', 'NHW', 'base_noapoe', 'age + sex + id_education + interview_language',
'cdr', 'NHW', 'base_noapoe', 'age + sex + id_education + interview_language',
'ab40', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'ab42', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'ab42_ab40', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'tau', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'ptau', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'nfl', 'NHW', 'base_noapoe', 'age + sex + bmi + eGFR',
'meta_roi', 'NHW', 'base_noapoe', 'age + sex + icv',
'z_hippcampul_vol', 'NHW', 'base_noapoe', 'age + sex + icv',
'wmh_volume_log', 'NHW', 'base_noapoe', 'age + sex + icv',
#mCAIDE
'ef', 'NHW', 'mcaide', 'interview_language + apoe + mcaide',
'em', 'NHW', 'mcaide', 'interview_language + apoe + mcaide',
'va', 'NHW', 'mcaide', 'interview_language + apoe + mcaide',
'mmse_total', 'NHW', 'mcaide', 'interview_language + apoe + mcaide',
'cdr', 'NHW', 'mcaide', 'interview_language + apoe + mcaide',
'ab40', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'ab42', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'ab42_ab40', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'tau', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'ptau', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'nfl', 'NHW', 'mcaide', 'apoe + eGFR + mcaide',
'meta_roi', 'NHW', 'mcaide', 'apoe + icv + mcaide',
'z_hippcampul_vol', 'NHW', 'mcaide', 'apoe + icv + mcaide',
'wmh_volume_log', 'NHW', 'mcaide', 'apoe + icv + mcaide',
# LIBRA
'ef', 'NHW', 'libra', 'interview_language + apoe + libra',
'em', 'NHW', 'libra', 'interview_language + apoe + libra',
'va', 'NHW', 'libra', 'interview_language + apoe + libra',
'mmse_total', 'NHW', 'libra', 'interview_language + apoe + libra',
'cdr', 'NHW', 'libra', 'interview_language + apoe + libra',
'ab40', 'NHW', 'libra', 'apoe + libra',
'ab42', 'NHW', 'libra', 'apoe + libra',
'ab42_ab40', 'NHW', 'libra', 'apoe + libra',
'tau', 'NHW', 'libra', 'apoe + libra',
'ptau', 'NHW', 'libra', 'apoe + libra',
'nfl', 'NHW', 'libra', 'apoe + libra',
'meta_roi', 'NHW', 'libra', 'apoe + icv + libra',
'z_hippcampul_vol', 'NHW', 'libra', 'apoe + icv + libra',
'wmh_volume_log', 'NHW', 'libra', 'apoe + icv + libra',
# WHICAP
'ef', 'NHW', 'whicap', 'interview_language + apoe + whicap',
'em', 'NHW', 'whicap', 'interview_language + apoe + whicap',
'va', 'NHW', 'whicap', 'interview_language + apoe + whicap',
'mmse_total', 'NHW', 'whicap', 'interview_language + apoe + whicap',
'cdr', 'NHW', 'whicap', 'interview_language + apoe + whicap',
'ab40', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'ab42', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'ab42_ab40', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'tau', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'ptau', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'nfl', 'NHW', 'whicap', 'apoe + eGFR + whicap',
'meta_roi', 'NHW', 'whicap', 'apoe + icv + whicap',
'z_hippcampul_vol', 'NHW', 'whicap', 'apoe + icv + whicap',
'wmh_volume_log', 'NHW', 'whicap', 'apoe + icv + whicap',
# CogD
'ef', 'NHW', 'cogd', 'interview_language + apoe + cogd',
'em', 'NHW', 'cogd', 'interview_language + apoe + cogd',
'va', 'NHW', 'cogd', 'interview_language + apoe + cogd',
'mmse_total', 'NHW', 'cogd', 'interview_language + apoe + cogd',
'cdr', 'NHW', 'cogd', 'interview_language + apoe + cogd',
'ab40', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42_ab40', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'tau', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'ptau', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'nfl', 'NHW', 'cogd', 'apoe + bmi + eGFR + cogd',
'meta_roi', 'NHW', 'cogd', 'apoe + icv + cogd',
'z_hippcampul_vol', 'NHW', 'cogd', 'apoe + icv + cogd',
'wmh_volume_log', 'NHW', 'cogd', 'apoe + icv + cogd',
# Sensitivity analysis
#LIBRA
'ef', 'NHW', 'libra_sva', 'interview_language + apoe + libra_sva',
'em', 'NHW', 'libra_sva', 'interview_language + apoe + libra_sva',
'va', 'NHW', 'libra_sva', 'interview_language + apoe + libra_sva',
'mmse_total', 'NHW', 'libra_sva', 'interview_language + apoe + libra_sva',
'cdr', 'NHW', 'libra_sva', 'interview_language + apoe + libra_sva',
'ab40', 'NHW', 'libra_sva', 'apoe + libra_sva',
'ab42', 'NHW', 'libra_sva', 'apoe + libra_sva',
'ab42_ab40', 'NHW', 'libra_sva', 'apoe + libra_sva',
'tau', 'NHW', 'libra_sva', 'apoe + libra_sva',
'ptau', 'NHW', 'libra_sva', 'apoe + libra_sva',
'nfl', 'NHW', 'libra_sva', 'apoe + libra_sva',
'meta_roi', 'NHW', 'libra_sva', 'apoe + icv + libra_sva',
'z_hippcampul_vol', 'NHW', 'libra_sva', 'apoe + icv + libra_sva',
'wmh_volume_log', 'NHW', 'libra_sva', 'apoe + icv + libra_sva',
#mCAIDE
'ef', 'NHW', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'em', 'NHW', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'va', 'NHW', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'mmse_total', 'NHW', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'cdr', 'NHW', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'ab40', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42_ab40', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'tau', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ptau', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'nfl', 'NHW', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'meta_roi', 'NHW', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'z_hippcampul_vol', 'NHW', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'wmh_volume_log', 'NHW', 'mcaide_sva', 'apoe + icv + mcaide_sva',
#WHICAP
'ef', 'NHW', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'em', 'NHW', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'va', 'NHW', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'mmse_total', 'NHW', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'cdr', 'NHW', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'ab40', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42_ab40', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'tau', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ptau', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'nfl', 'NHW', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'meta_roi', 'NHW', 'whicap_sva', 'apoe + icv + whicap_sva',
'z_hippcampul_vol', 'NHW', 'whicap_sva', 'apoe + icv + whicap_sva',
'wmh_volume_log', 'NHW', 'whicap_sva', 'apoe + icv + whicap_sva',
#CogD
'ef', 'NHW', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'em', 'NHW', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'va', 'NHW', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'mmse_total', 'NHW', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'cdr', 'NHW', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'ab40', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42_ab40', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'tau', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ptau', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'nfl', 'NHW', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'meta_roi', 'NHW', 'cogd_sva', 'apoe + icv + cogd_sva',
'z_hippcampul_vol', 'NHW', 'cogd_sva', 'apoe + icv + cogd_sva',
'wmh_volume_log', 'NHW', 'cogd_sva', 'apoe + icv + cogd_sva'
) %>%
mutate(eq = glue('{outcome} ~ {predictors}'),
res = map(eq, lm, data = filter(data, race == "NHW")),
dataf = map(res, tidy),
mod = map(res, glance),
r2 = map_dbl(mod, ~ .x$r.squared),
r2.adj = map_dbl(mod, ~ .x$adj.r.squared),
n = map_int(res, nobs))
amr_num_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'ef', 'Hispanic', 'base', 'interview_language + age + sex + id_education + apoe',
'em', 'Hispanic', 'base', 'interview_language + age + sex + id_education + apoe',
'va', 'Hispanic', 'base', 'interview_language + age + sex + id_education + apoe',
'mmse_total', 'Hispanic', 'base', 'apoe + age + sex + id_education + interview_language',
'cdr', 'Hispanic', 'base', 'apoe + age + sex + id_education + interview_language',
'ab40', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42_ab40', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'tau', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'ptau', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'nfl', 'Hispanic', 'base', 'apoe + age + sex + bmi + eGFR',
'meta_roi', 'Hispanic', 'base', 'apoe + age + sex + icv',
'z_hippcampul_vol', 'Hispanic', 'base', 'apoe + age + sex + icv',
'wmh_volume_log', 'Hispanic', 'base', 'apoe + age + sex + icv',
#Demographics - apoe
'ef', 'Hispanic', 'noapoe', 'interview_language + age + sex + id_education',
'em', 'Hispanic', 'noapoe', 'interview_language + age + sex + id_education',
'va', 'Hispanic', 'noapoe', 'interview_language + age + sex + id_education',
'mmse_total', 'Hispanic', 'noapoe', 'age + sex + id_education + interview_language',
'cdr', 'Hispanic', 'noapoe', 'age + sex + id_education + interview_language',
'ab40', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'ab42', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'ab42_ab40', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'tau', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'ptau', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'nfl', 'Hispanic', 'noapoe', 'age + sex + bmi + eGFR',
'meta_roi', 'Hispanic', 'noapoe', 'age + sex + icv',
'z_hippcampul_vol', 'Hispanic', 'noapoe', 'age + sex + icv',
'wmh_volume_log', 'Hispanic', 'noapoe', 'age + sex + icv',
#mCAIDE
'ef', 'Hispanic', 'mcaide', 'interview_language + apoe + mcaide',
'em', 'Hispanic', 'mcaide', 'interview_language + apoe + mcaide',
'va', 'Hispanic', 'mcaide', 'interview_language + apoe + mcaide',
'mmse_total', 'Hispanic', 'mcaide', 'interview_language + apoe + mcaide',
'cdr', 'Hispanic', 'mcaide', 'interview_language + apoe + mcaide',
'ab40', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'ab42', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'ab42_ab40', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'tau', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'ptau', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'nfl', 'Hispanic', 'mcaide', 'apoe + eGFR + mcaide',
'meta_roi', 'Hispanic', 'mcaide', 'apoe + icv + mcaide',
'z_hippcampul_vol', 'Hispanic', 'mcaide', 'apoe + icv + mcaide',
'wmh_volume_log', 'Hispanic', 'mcaide', 'apoe + icv + mcaide',
# LIBRA
'ef', 'Hispanic', 'libra', 'interview_language + apoe + libra',
'em', 'Hispanic', 'libra', 'interview_language + apoe + libra',
'va', 'Hispanic', 'libra', 'interview_language + apoe + libra',
'mmse_total', 'Hispanic', 'libra', 'interview_language + apoe + libra',
'cdr', 'Hispanic', 'libra', 'interview_language + apoe + libra',
'ab40', 'Hispanic', 'libra', 'apoe + libra',
'ab42', 'Hispanic', 'libra', 'apoe + libra',
'ab42_ab40', 'Hispanic', 'libra', 'apoe + libra',
'tau', 'Hispanic', 'libra', 'apoe + libra',
'ptau', 'Hispanic', 'libra', 'apoe + libra',
'nfl', 'Hispanic', 'libra', 'apoe + libra',
'meta_roi', 'Hispanic', 'libra', 'apoe + icv + libra',
'z_hippcampul_vol', 'Hispanic', 'libra', 'apoe + icv + libra',
'wmh_volume_log', 'Hispanic', 'libra', 'apoe + icv + libra',
# WHICAP
'ef', 'Hispanic', 'whicap', 'interview_language + apoe + whicap',
'em', 'Hispanic', 'whicap', 'interview_language + apoe + whicap',
'va', 'Hispanic', 'whicap', 'interview_language + apoe + whicap',
'mmse_total', 'Hispanic', 'whicap', 'interview_language + apoe + whicap',
'cdr', 'Hispanic', 'whicap', 'interview_language + apoe + whicap',
'ab40', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'ab42', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'ab42_ab40', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'tau', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'ptau', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'nfl', 'Hispanic', 'whicap', 'apoe + eGFR + whicap',
'meta_roi', 'Hispanic', 'whicap', 'apoe + icv + whicap',
'z_hippcampul_vol', 'Hispanic', 'whicap', 'apoe + icv + whicap',
'wmh_volume_log', 'Hispanic', 'whicap', 'apoe + icv + whicap',
# CogD
'ef', 'Hispanic', 'cogd', 'interview_language + apoe + cogd',
'em', 'Hispanic', 'cogd', 'interview_language + apoe + cogd',
'va', 'Hispanic', 'cogd', 'interview_language + apoe + cogd',
'mmse_total', 'Hispanic', 'cogd', 'interview_language + apoe + cogd',
'cdr', 'Hispanic', 'cogd', 'interview_language + apoe + cogd',
'ab40', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42_ab40', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'tau', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'ptau', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'nfl', 'Hispanic', 'cogd', 'apoe + bmi + eGFR + cogd',
'meta_roi', 'Hispanic', 'cogd', 'apoe + icv + cogd',
'z_hippcampul_vol', 'Hispanic', 'cogd', 'apoe + icv + cogd',
'wmh_volume_log', 'Hispanic', 'cogd', 'apoe + icv + cogd',
# Sensitivity analysis
#LIBRA
'ef', 'Hispanic', 'libra_sva', 'interview_language + apoe + libra_sva',
'em', 'Hispanic', 'libra_sva', 'interview_language + apoe + libra_sva',
'va', 'Hispanic', 'libra_sva', 'interview_language + apoe + libra_sva',
'mmse_total', 'Hispanic', 'libra_sva', 'interview_language + apoe + libra_sva',
'cdr', 'Hispanic', 'libra_sva', 'interview_language + apoe + libra_sva',
'ab40', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'ab42', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'ab42_ab40', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'tau', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'ptau', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'nfl', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'meta_roi', 'Hispanic', 'libra_sva', 'apoe + icv + libra_sva',
'z_hippcampul_vol', 'Hispanic', 'libra_sva', 'apoe + icv + libra_sva',
'wmh_volume_log', 'Hispanic', 'libra_sva', 'apoe + icv + libra_sva',
#mCAIDE
'ef', 'Hispanic', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'em', 'Hispanic', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'va', 'Hispanic', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'mmse_total', 'Hispanic', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'cdr', 'Hispanic', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'ab40', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42_ab40', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'tau', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ptau', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'nfl', 'Hispanic', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'meta_roi', 'Hispanic', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'z_hippcampul_vol', 'Hispanic', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'wmh_volume_log', 'Hispanic', 'mcaide_sva', 'apoe + icv + mcaide_sva',
#WHICAP
'ef', 'Hispanic', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'em', 'Hispanic', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'va', 'Hispanic', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'mmse_total', 'Hispanic', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'cdr', 'Hispanic', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'ab40', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42_ab40', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'tau', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ptau', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'nfl', 'Hispanic', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'meta_roi', 'Hispanic', 'whicap_sva', 'apoe + icv + whicap_sva',
'z_hippcampul_vol', 'Hispanic', 'whicap_sva', 'apoe + icv + whicap_sva',
'wmh_volume_log', 'Hispanic', 'whicap_sva', 'apoe + icv + whicap_sva',
#CogD
'ef', 'Hispanic', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'em', 'Hispanic', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'va', 'Hispanic', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'mmse_total', 'Hispanic', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'cdr', 'Hispanic', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'ab40', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42_ab40', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'tau', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ptau', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'nfl', 'Hispanic', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'meta_roi', 'Hispanic', 'cogd_sva', 'apoe + icv + cogd_sva',
'z_hippcampul_vol', 'Hispanic', 'cogd_sva', 'apoe + icv + cogd_sva',
'wmh_volume_log', 'Hispanic', 'cogd_sva', 'apoe + icv + cogd_sva'
) %>%
mutate(eq = glue('{outcome} ~ {predictors}'),
res = map(eq, lm, data = filter(data, race == "Hispanic")),
dataf = map(res, tidy),
mod = map(res, glance),
r2 = map_dbl(mod, ~ .x$r.squared),
r2.adj = map_dbl(mod, ~ .x$adj.r.squared),
n = map_int(res, nobs))
afr_num_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'ef', 'Black', 'base', 'interview_language + age + sex + id_education + apoe',
'em', 'Black', 'base', 'interview_language + age + sex + id_education + apoe',
'va', 'Black', 'base', 'interview_language + age + sex + id_education + apoe',
'mmse_total', 'Black', 'base', 'apoe + age + sex + id_education + interview_language',
'cdr', 'Black', 'base', 'apoe + age + sex + id_education + interview_language',
'ab40', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'ab42_ab40', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'tau', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'ptau', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'nfl', 'Black', 'base', 'apoe + age + sex + bmi + eGFR',
'meta_roi', 'Black', 'base', 'apoe + age + sex + icv',
'z_hippcampul_vol', 'Black', 'base', 'apoe + age + sex + icv',
'wmh_volume_log', 'Black', 'base', 'apoe + age + sex + icv',
#Demographics - apoe
'ef', 'Black', 'noapoe', 'interview_language + age + sex + id_education',
'em', 'Black', 'noapoe', 'interview_language + age + sex + id_education',
'va', 'Black', 'noapoe', 'interview_language + age + sex + id_education',
'mmse_total', 'Black', 'noapoe', 'age + sex + id_education + interview_language',
'cdr', 'Black', 'noapoe', 'age + sex + id_education + interview_language',
'ab40', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'ab42', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'ab42_ab40', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'tau', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'ptau', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'nfl', 'Black', 'noapoe', 'age + sex + bmi + eGFR',
'meta_roi', 'Black', 'noapoe', 'age + sex + icv',
'z_hippcampul_vol', 'Black', 'noapoe', 'age + sex + icv',
'wmh_volume_log', 'Black', 'noapoe', 'age + sex + icv',
#mCAIDE
'ef', 'Black', 'mcaide', 'interview_language + apoe + mcaide',
'em', 'Black', 'mcaide', 'interview_language + apoe + mcaide',
'va', 'Black', 'mcaide', 'interview_language + apoe + mcaide',
'mmse_total', 'Black', 'mcaide', 'interview_language + apoe + mcaide',
'cdr', 'Black', 'mcaide', 'interview_language + apoe + mcaide',
'ab40', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'ab42', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'ab42_ab40', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'tau', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'ptau', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'nfl', 'Black', 'mcaide', 'apoe + eGFR + mcaide',
'meta_roi', 'Black', 'mcaide', 'apoe + icv + mcaide',
'z_hippcampul_vol', 'Black', 'mcaide', 'apoe + icv + mcaide',
'wmh_volume_log', 'Black', 'mcaide', 'apoe + icv + mcaide',
# LIBRA
'ef', 'Black', 'libra', 'interview_language + apoe + libra',
'em', 'Black', 'libra', 'interview_language + apoe + libra',
'va', 'Black', 'libra', 'interview_language + apoe + libra',
'mmse_total', 'Black', 'libra', 'interview_language + apoe + libra',
'cdr', 'Black', 'libra', 'interview_language + apoe + libra',
'ab40', 'Black', 'libra', 'apoe + libra',
'ab42', 'Black', 'libra', 'apoe + libra',
'ab42_ab40', 'Black', 'libra', 'apoe + libra',
'tau', 'Black', 'libra', 'apoe + libra',
'ptau', 'Black', 'libra', 'apoe + libra',
'nfl', 'Black', 'libra', 'apoe + libra',
'meta_roi', 'Black', 'libra', 'apoe + icv + libra',
'z_hippcampul_vol', 'Black', 'libra', 'apoe + icv + libra',
'wmh_volume_log', 'Black', 'libra', 'apoe + icv + libra',
# WHICAP
'ef', 'Black', 'whicap', 'interview_language + apoe + whicap',
'em', 'Black', 'whicap', 'interview_language + apoe + whicap',
'va', 'Black', 'whicap', 'interview_language + apoe + whicap',
'mmse_total', 'Black', 'whicap', 'interview_language + apoe + whicap',
'cdr', 'Black', 'whicap', 'interview_language + apoe + whicap',
'ab40', 'Black', 'whicap', 'apoe + eGFR + whicap',
'ab42', 'Black', 'whicap', 'apoe + eGFR + whicap',
'ab42_ab40', 'Black', 'whicap', 'apoe + eGFR + whicap',
'tau', 'Black', 'whicap', 'apoe + eGFR + whicap',
'ptau', 'Black', 'whicap', 'apoe + eGFR + whicap',
'nfl', 'Black', 'whicap', 'apoe + eGFR + whicap',
'meta_roi', 'Black', 'whicap', 'apoe + icv + whicap',
'z_hippcampul_vol', 'Black', 'whicap', 'apoe + icv + whicap',
'wmh_volume_log', 'Black', 'whicap', 'apoe + icv + whicap',
# CogD
'ef', 'Black', 'cogd', 'interview_language + apoe + cogd',
'em', 'Black', 'cogd', 'interview_language + apoe + cogd',
'va', 'Black', 'cogd', 'interview_language + apoe + cogd',
'mmse_total', 'Black', 'cogd', 'interview_language + apoe + cogd',
'cdr', 'Black', 'cogd', 'interview_language + apoe + cogd',
'ab40', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'ab42_ab40', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'tau', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'ptau', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'nfl', 'Black', 'cogd', 'apoe + bmi + eGFR + cogd',
'meta_roi', 'Black', 'cogd', 'apoe + icv + cogd',
'z_hippcampul_vol', 'Black', 'cogd', 'apoe + icv + cogd',
'wmh_volume_log', 'Black', 'cogd', 'apoe + icv + cogd',
# Sensitivity analysis
#LIBRA
'ef', 'Black', 'libra_sva', 'interview_language + apoe + libra_sva',
'em', 'Black', 'libra_sva', 'interview_language + apoe + libra_sva',
'va', 'Black', 'libra_sva', 'interview_language + apoe + libra_sva',
'mmse_total', 'Black', 'libra_sva', 'interview_language + apoe + libra_sva',
'cdr', 'Black', 'libra_sva', 'interview_language + apoe + libra_sva',
'ab40', 'Black', 'libra_sva', 'apoe + libra_sva',
'ab42', 'Black', 'libra_sva', 'apoe + libra_sva',
'ab42_ab40', 'Black', 'libra_sva', 'apoe + libra_sva',
'tau', 'Black', 'libra_sva', 'apoe + libra_sva',
'ptau', 'Black', 'libra_sva', 'apoe + libra_sva',
'nfl', 'Black', 'libra_sva', 'apoe + libra_sva',
'meta_roi', 'Black', 'libra_sva', 'apoe + icv + libra_sva',
'z_hippcampul_vol', 'Black', 'libra_sva', 'apoe + icv + libra_sva',
'wmh_volume_log', 'Black', 'libra_sva', 'apoe + icv + libra_sva',
#mCAIDE
'ef', 'Black', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'em', 'Black', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'va', 'Black', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'mmse_total', 'Black', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'cdr', 'Black', 'mcaide_sva', 'interview_language + apoe + mcaide_sva',
'ab40', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ab42_ab40', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'tau', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'ptau', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'nfl', 'Black', 'mcaide_sva', 'apoe + eGFR + mcaide_sva',
'meta_roi', 'Black', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'z_hippcampul_vol', 'Black', 'mcaide_sva', 'apoe + icv + mcaide_sva',
'wmh_volume_log', 'Black', 'mcaide_sva', 'apoe + icv + mcaide_sva',
#WHICAP
'ef', 'Black', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'em', 'Black', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'va', 'Black', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'mmse_total', 'Black', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'cdr', 'Black', 'whicap_sva', 'interview_language + apoe + whicap_sva',
'ab40', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ab42_ab40', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'tau', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'ptau', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'nfl', 'Black', 'whicap_sva', 'apoe + eGFR + whicap_sva',
'meta_roi', 'Black', 'whicap_sva', 'apoe + icv + whicap_sva',
'z_hippcampul_vol', 'Black', 'whicap_sva', 'apoe + icv + whicap_sva',
'wmh_volume_log', 'Black', 'whicap_sva', 'apoe + icv + whicap_sva',
#CogD
'ef', 'Black', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'em', 'Black', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'va', 'Black', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'mmse_total', 'Black', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'cdr', 'Black', 'cogd_sva', 'interview_language + apoe + cogd_sva',
'ab40', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ab42_ab40', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'tau', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'ptau', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'nfl', 'Black', 'cogd_sva', 'apoe + eGFR + cogd_sva',
'meta_roi', 'Black', 'cogd_sva', 'apoe + icv + cogd_sva',
'z_hippcampul_vol', 'Black', 'cogd_sva', 'apoe + icv + cogd_sva',
'wmh_volume_log', 'Black', 'cogd_sva', 'apoe + icv + cogd_sva'
) %>%
mutate(eq = glue('{outcome} ~ {predictors}'),
res = map(eq, lm, data = filter(data, race == "Black")),
dataf = map(res, tidy),
mod = map(res, glance),
r2 = map_dbl(mod, ~ .x$r.squared),
r2.adj = map_dbl(mod, ~ .x$adj.r.squared),
n = map_int(res, nobs))
```
### CDX - Ancestry
```{r}
eur_cdx_prs = tribble( ~outcome, ~pop, ~model, ~predictors,
#Demographics
'cdx_cn_vs_ci', 'NHW', 'base', 'age + sex + apoe + id_education',
'cdx_mci_vs_cn', 'NHW', 'base', 'age + sex + apoe + id_education',
'cdx_dem_vs_cn', 'NHW', 'base', 'age + sex + apoe + id_education',
'cdx_ci_vs_cn', 'NHW', 'base', 'age + sex + apoe + id_education',
#Demographics
'cdx_cn_vs_ci', 'NHW', 'base_noapoe', 'age + sex + id_education',
'cdx_mci_vs_cn', 'NHW', 'base_noapoe', 'age + sex + id_education',
'cdx_dem_vs_cn', 'NHW', 'base_noapoe', 'age + sex + id_education',
'cdx_ci_vs_cn', 'NHW', 'base_noapoe', 'age + sex + id_education',
#LIBRA
'cdx_cn_vs_ci', 'NHW', 'libra', 'apoe + libra',
'cdx_mci_vs_cn', 'NHW', 'libra', 'apoe + libra',
'cdx_dem_vs_cn', 'NHW', 'libra', 'apoe + libra',
'cdx_ci_vs_cn', 'NHW', 'libra', 'apoe + libra',
#mCAIDE
'cdx_cn_vs_ci', 'NHW', 'mcaide', 'apoe + mcaide',
'cdx_mci_vs_cn', 'NHW', 'mcaide', 'apoe + mcaide',
'cdx_dem_vs_cn', 'NHW', 'mcaide', 'apoe + mcaide',
'cdx_ci_vs_cn', 'NHW', 'mcaide', 'apoe + mcaide',
#WHICAP
'cdx_cn_vs_ci', 'NHW', 'whicap', 'apoe + whicap',
'cdx_mci_vs_cn', 'NHW', 'whicap', 'apoe + whicap',
'cdx_dem_vs_cn', 'NHW', 'whicap', 'apoe + whicap',
'cdx_ci_vs_cn', 'NHW', 'whicap', 'apoe + whicap',
#CogD
'cdx_cn_vs_ci', 'NHW', 'cogd', 'apoe + cogd',
'cdx_mci_vs_cn', 'NHW', 'cogd', 'apoe + cogd',
'cdx_dem_vs_cn', 'NHW', 'cogd', 'apoe + cogd',
'cdx_ci_vs_cn', 'NHW', 'cogd', 'apoe + cogd',
# Sensitivity analysis
#LIBRA
'cdx_cn_vs_ci', 'NHW', 'libra_sva', 'apoe + libra_sva',
'cdx_mci_vs_cn', 'NHW', 'libra_sva', 'apoe + libra_sva',
'cdx_dem_vs_cn', 'NHW', 'libra_sva', 'apoe + libra_sva',
'cdx_ci_vs_cn', 'NHW', 'libra_sva', 'apoe + libra_sva',
#mCAIDE
'cdx_cn_vs_ci', 'NHW', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_mci_vs_cn', 'NHW', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_dem_vs_cn', 'NHW', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_ci_vs_cn', 'NHW', 'mcaide_sva', 'apoe + mcaide_sva',
#WHICAP
'cdx_cn_vs_ci', 'NHW', 'whicap_sva', 'apoe + whicap_sva',
'cdx_mci_vs_cn', 'NHW', 'whicap_sva', 'apoe + whicap_sva',
'cdx_dem_vs_cn', 'NHW', 'whicap_sva', 'apoe + whicap_sva',
'cdx_ci_vs_cn', 'NHW', 'whicap_sva', 'apoe + whicap_sva',
#CogD
'cdx_cn_vs_ci', 'NHW', 'cogd_sva', 'apoe + cogd_sva',
'cdx_mci_vs_cn', 'NHW', 'cogd_sva', 'apoe + cogd_sva',
'cdx_dem_vs_cn', 'NHW', 'cogd_sva', 'apoe + cogd_sva',
'cdx_ci_vs_cn', 'NHW', 'cogd_sva', 'apoe + cogd_sva'
) %>%
mutate(
eq = glue("{outcome} ~ {predictors}"),
res = map(eq, glm, data = filter(data, race == "NHW"), family = "binomial"),
auc = map_dbl(res, ~ {
y = model.response(model.frame(.x)) # create model.frame(x) which is the exact same data as res
p = predict(.x, type = "response") # predict fitted probabilities for each outcome
if (length(unique(y)) > 1) { # if outcome is all 0s or 1s, skip
as.numeric(pROC::auc(pROC::roc(y, p))) # compute ROC curve then AUC and return numeric number (as.numeric)
} else {NA_real_}}), # if the outcome has only one class, return NA
nagelkerke_r2 = map_dbl(res, ~ {
out = tryCatch(performance::r2_nagelkerke(.x), error = function(e) NA_real_)
as.numeric(out)}),
dataf = map(res, tidy),
mod = map(res, glance),
n = map_int(res, nobs))
amr_cdx_prs = tribble( ~outcome, ~pop, ~model, ~predictors,
#Demographics
'cdx_cn_vs_ci', 'Hispanic', 'base', 'age + sex + apoe + id_education',
'cdx_mci_vs_cn', 'Hispanic', 'base', 'age + sex + apoe + id_education',
'cdx_dem_vs_cn', 'Hispanic', 'base', 'age + sex + apoe + id_education',
'cdx_ci_vs_cn', 'Hispanic', 'base', 'age + sex + apoe + id_education',
#Demographics
'cdx_cn_vs_ci', 'Hispanic', 'base_noapoe', 'age + sex + id_education',
'cdx_mci_vs_cn', 'Hispanic', 'base_noapoe', 'age + sex + id_education',
'cdx_dem_vs_cn', 'Hispanic', 'base_noapoe', 'age + sex + id_education',
'cdx_ci_vs_cn', 'Hispanic', 'base_noapoe', 'age + sex + id_education',
#LIBRA
'cdx_cn_vs_ci', 'Hispanic', 'libra', 'apoe + libra',
'cdx_mci_vs_cn', 'Hispanic', 'libra', 'apoe + libra',
'cdx_dem_vs_cn', 'Hispanic', 'libra', 'apoe + libra',
'cdx_ci_vs_cn', 'Hispanic', 'libra', 'apoe + libra',
#mCAIDE
'cdx_cn_vs_ci', 'Hispanic', 'mcaide', 'apoe + mcaide',
'cdx_mci_vs_cn', 'Hispanic', 'mcaide', 'apoe + mcaide',
'cdx_dem_vs_cn', 'Hispanic', 'mcaide', 'apoe + mcaide',
'cdx_ci_vs_cn', 'Hispanic', 'mcaide', 'apoe + mcaide',
#WHICAP
'cdx_cn_vs_ci', 'Hispanic', 'whicap', 'apoe + whicap',
'cdx_mci_vs_cn', 'Hispanic', 'whicap', 'apoe + whicap',
'cdx_dem_vs_cn', 'Hispanic', 'whicap', 'apoe + whicap',
'cdx_ci_vs_cn', 'Hispanic', 'whicap', 'apoe + whicap',
#CogD
'cdx_cn_vs_ci', 'Hispanic', 'cogd', 'apoe + cogd',
'cdx_mci_vs_cn', 'Hispanic', 'cogd', 'apoe + cogd',
'cdx_dem_vs_cn', 'Hispanic', 'cogd', 'apoe + cogd',
'cdx_ci_vs_cn', 'Hispanic', 'cogd', 'apoe + cogd',
# Sensitivity analysis
#LIBRA
'cdx_cn_vs_ci', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'cdx_mci_vs_cn', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'cdx_dem_vs_cn', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
'cdx_ci_vs_cn', 'Hispanic', 'libra_sva', 'apoe + libra_sva',
#mCAIDE
'cdx_cn_vs_ci', 'Hispanic', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_mci_vs_cn', 'Hispanic', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_dem_vs_cn', 'Hispanic', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_ci_vs_cn', 'Hispanic', 'mcaide_sva', 'apoe + mcaide_sva',
#WHICAP
'cdx_cn_vs_ci', 'Hispanic', 'whicap_sva', 'apoe + whicap_sva',
'cdx_mci_vs_cn', 'Hispanic', 'whicap_sva', 'apoe + whicap_sva',
'cdx_dem_vs_cn', 'Hispanic', 'whicap_sva', 'apoe + whicap_sva',
'cdx_ci_vs_cn', 'Hispanic', 'whicap_sva', 'apoe + whicap_sva',
#CogD
'cdx_cn_vs_ci', 'Hispanic', 'cogd_sva', 'apoe + cogd_sva',
'cdx_mci_vs_cn', 'Hispanic', 'cogd_sva', 'apoe + cogd_sva',
'cdx_dem_vs_cn', 'Hispanic', 'cogd_sva', 'apoe + cogd_sva',
'cdx_ci_vs_cn', 'Hispanic', 'cogd_sva', 'apoe + cogd_sva'
) %>%
mutate(
eq = glue("{outcome} ~ {predictors}"),
res = map(eq, glm, data = filter(data, race == "Hispanic"), family = "binomial"),
auc = map_dbl(res, ~ {
y = model.response(model.frame(.x)) # create model.frame(x) which is the exact same data as res
p = predict(.x, type = "response") # predict fitted probabilities for each outcome
if (length(unique(y)) > 1) { # if outcome is all 0s or 1s, skip
as.numeric(pROC::auc(pROC::roc(y, p))) # compute ROC curve then AUC and return numeric number (as.numeric)
} else {NA_real_}}), # if the outcome has only one class, return NA
nagelkerke_r2 = map_dbl(res, ~ {
out = tryCatch(performance::r2_nagelkerke(.x), error = function(e) NA_real_)
as.numeric(out)}),
dataf = map(res, tidy),
mod = map(res, glance),
n = map_int(res, nobs))
afr_cdx_prs = tribble(~outcome, ~pop, ~model, ~predictors,
#Demographics
'cdx_cn_vs_ci', 'Black', 'base', 'age + sex + apoe + id_education',
'cdx_mci_vs_cn', 'Black', 'base', 'age + sex + apoe + id_education',
'cdx_dem_vs_cn', 'Black', 'base', 'age + sex + apoe + id_education',
'cdx_ci_vs_cn', 'Black', 'base', 'age + sex + apoe + id_education',
#Demographics
'cdx_cn_vs_ci', 'Black', 'base_noapoe', 'age + sex + id_education',
'cdx_mci_vs_cn', 'Black', 'base_noapoe', 'age + sex + id_education',
'cdx_dem_vs_cn', 'Black', 'base_noapoe', 'age + sex + id_education',
'cdx_ci_vs_cn', 'Black', 'base_noapoe', 'age + sex + id_education',
#LIBRA
'cdx_cn_vs_ci', 'Black', 'libra', 'apoe + libra',
'cdx_mci_vs_cn', 'Black', 'libra', 'apoe + libra',
'cdx_dem_vs_cn', 'Black', 'libra', 'apoe + libra',
'cdx_ci_vs_cn', 'Black', 'libra', 'apoe + libra',
#mCAIDE
'cdx_cn_vs_ci', 'Black', 'mcaide', 'apoe + mcaide',
'cdx_mci_vs_cn', 'Black', 'mcaide', 'apoe + mcaide',
'cdx_dem_vs_cn', 'Black', 'mcaide', 'apoe + mcaide',
'cdx_ci_vs_cn', 'Black', 'mcaide', 'apoe + mcaide',
#WHICAP
'cdx_cn_vs_ci', 'Black', 'whicap', 'apoe + whicap',
'cdx_mci_vs_cn', 'Black', 'whicap', 'apoe + whicap',
'cdx_dem_vs_cn', 'Black', 'whicap', 'apoe + whicap',
'cdx_ci_vs_cn', 'Black', 'whicap', 'apoe + whicap',
#CogD
'cdx_cn_vs_ci', 'Black', 'cogd', 'apoe + cogd',
'cdx_mci_vs_cn', 'Black', 'cogd', 'apoe + cogd',
'cdx_dem_vs_cn', 'Black', 'cogd', 'apoe + cogd',
'cdx_ci_vs_cn', 'Black', 'cogd', 'apoe + cogd',
# Sensitivity analysis
#LIBRA
'cdx_cn_vs_ci', 'Black', 'libra_sva', 'apoe + libra_sva',
'cdx_mci_vs_cn', 'Black', 'libra_sva', 'apoe + libra_sva',
'cdx_dem_vs_cn', 'Black', 'libra_sva', 'apoe + libra_sva',
'cdx_ci_vs_cn', 'Black', 'libra_sva', 'apoe + libra_sva',
#mCAIDE
'cdx_cn_vs_ci', 'Black', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_mci_vs_cn', 'Black', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_dem_vs_cn', 'Black', 'mcaide_sva', 'apoe + mcaide_sva',
'cdx_ci_vs_cn', 'Black', 'mcaide_sva', 'apoe + mcaide_sva',
#WHICAP
'cdx_cn_vs_ci', 'Black', 'whicap_sva', 'apoe + whicap_sva',
'cdx_mci_vs_cn', 'Black', 'whicap_sva', 'apoe + whicap_sva',
'cdx_dem_vs_cn', 'Black', 'whicap_sva', 'apoe + whicap_sva',
'cdx_ci_vs_cn', 'Black', 'whicap_sva', 'apoe + whicap_sva',
#CogD
'cdx_cn_vs_ci', 'Black', 'cogd_sva', 'apoe + cogd_sva',
'cdx_mci_vs_cn', 'Black', 'cogd_sva', 'apoe + cogd_sva',
'cdx_dem_vs_cn', 'Black', 'cogd_sva', 'apoe + cogd_sva',
'cdx_ci_vs_cn', 'Black', 'cogd_sva', 'apoe + cogd_sva'
) %>%
mutate(
eq = glue("{outcome} ~ {predictors}"),
res = map(eq, glm, data = filter(data, race == "Black"), family = "binomial"),
auc = map_dbl(res, ~ {
y = model.response(model.frame(.x)) # create model.frame(x) which is the exact same data as res
p = predict(.x, type = "response") # predict fitted probabilities for each outcome
if (length(unique(y)) > 1) { # if outcome is all 0s or 1s, skip
as.numeric(pROC::auc(pROC::roc(y, p))) # compute ROC curve then AUC and return numeric number (as.numeric)
} else {NA_real_}}), # if the outcome has only one class, return NA
nagelkerke_r2 = map_dbl(res, ~ {
out = tryCatch(performance::r2_nagelkerke(.x), error = function(e) NA_real_)
as.numeric(out)}),
dataf = map(res, tidy),
mod = map(res, glance),
n = map_int(res, nobs))
```
### Coefficients
```{r}
res_crs = bind_rows(
select(all_num_prs, model, pop, outcome, dataf, res),
select(all_cdx_prs, model, pop, outcome, dataf, res),
select(eur_num_prs, model, outcome, pop, dataf, res),
select(amr_num_prs, model, outcome, pop, dataf, res),
select(afr_num_prs, model, outcome, pop, dataf, res),
select(eur_cdx_prs, model, outcome, pop, dataf, res),
select(amr_cdx_prs, model, outcome, pop, dataf, res),
select(afr_cdx_prs, model, outcome, pop, dataf, res)
) %>%
unnest(dataf) %>%
mutate(std_params = map(res, ~ standardize_parameters(.x, method = "posthoc"))) %>%
unnest(std_params) %>%
select(-c(res, CI)) %>%
filter(term %in% c("cogd", "libra", "whicap", "mcaide")) %>%
filter(!outcome %in% c("cdx_hypertension", "cdx_dyslipidemia", "cdx_diabetes", "bmi", "cdx_anxiety", "cdx_depression", "cdx_cvd", "wmh_volume_log")) %>%
arrange(pop, term, Std_Coefficient) %>%
mutate(
p.fdr = p.adjust(p.value, method = "fdr"),
sig_fdr = ifelse(p.fdr < 0.05, TRUE, FALSE),
sig = ifelse(p.value < 0.05, TRUE, FALSE),
conf.low = estimate - (std.error * 1.96),
conf.high = estimate + (std.error * 1.96),
lci = exp(conf.low),
uci = exp(conf.high),
or = exp(estimate),
category = case_when(
outcome %in% c("va", "em", "ef", "mmse_total", "cdr") ~ "Cognitive \nFunction",
outcome %in% c("tau", "ptau", "nfl", "ab42", "ab40", "ab42_ab40") ~ "Plasma \nBiomarkers",
outcome %in% c("meta_roi", "z_hippcampul_vol") ~ "Neuroimaging",
outcome %in% c("cdx_cn_vs_ci", "cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn") ~ "Medical \nHistory"),
lab = fct_recode(
outcome,
MMSE = "mmse_total",
CDR = "cdr",
"Executive\n Function" = "ef",
"Verbal\n Ability" = "va",
Memory = "em",
"Cortical \nThickness" = "meta_roi",
"Hippo. \nVolume" = "z_hippcampul_vol",
"Aβ40" = "ab40",
"Aβ42" = "ab42",
"Aβ42/Aβ40" = "ab42_ab40",
"Total Tau" = "tau",
"pTau" = "ptau",
NfL = "nfl",
Healthy = "cdx_cn_vs_ci",
MCI = "cdx_mci_vs_cn",
Dementia = "cdx_dem_vs_cn",
"Cog. Impair." = "cdx_ci_vs_cn"
),
lab = fct_inorder(lab))
res_crs_sva = bind_rows(
select(all_num_prs, model, pop, outcome, dataf, res),
select(all_cdx_prs, model, pop, outcome, dataf, res),
select(eur_num_prs, model, outcome, pop, dataf, res),
select(amr_num_prs, model, outcome, pop, dataf, res),
select(afr_num_prs, model, outcome, pop, dataf, res),
select(eur_cdx_prs, model, outcome, pop, dataf, res),
select(amr_cdx_prs, model, outcome, pop, dataf, res),
select(afr_cdx_prs, model, outcome, pop, dataf, res)
) %>%
unnest(dataf) %>%
mutate(std_params = map(res, ~ standardize_parameters(.x, method = "posthoc"))) %>%
unnest(std_params) %>%
select(-c(res, CI)) %>%
filter(term %in% c("cogd_sva", "whicap_sva", "mcaide_sva")) %>%
filter(!outcome %in% c("cdx_hypertension", "cdx_dyslipidemia", "cdx_diabetes", "bmi", "cdx_anxiety", "cdx_depression", "cdx_cvd", "wmh_volume_log")) %>%
arrange(pop, term, estimate) %>%
mutate(
p.fdr = p.adjust(p.value, method = "fdr"),
sig_fdr = ifelse(p.fdr < 0.05, TRUE, FALSE),
sig = ifelse(p.value < 0.05, TRUE, FALSE),
conf.low = estimate - (std.error * 1.96),
conf.high = estimate + (std.error * 1.96),
lci = exp(conf.low),
uci = exp(conf.high),
or = exp(estimate),
category = case_when(
outcome %in% c("va", "em", "ef", "mmse_total", "cdr") ~ "Cognitive \nFunction",
outcome %in% c("tau", "ptau", "nfl", "ab42", "ab40", "ab42_ab40") ~ "Plasma \nBiomarkers",
outcome %in% c("meta_roi", "z_hippcampul_vol") ~ "Neuroimaging",
outcome %in% c("cdx_cn_vs_ci", "cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn") ~ "Medical \nHistory"),
outcome = factor(outcome, levels = c(
"cdx_hypertension", "cdx_dyslipidemia", "cdx_diabetes", "cdx_cvd",
"cdx_anxiety", "cdx_depression", "bmi",
"mmse_total", "cdr", "ef", "va", "em",
"meta_roi", "z_hippcampul_vol",
"ab40", "ab42", "ab42_ab40", "tau", "ptau", "nfl",
"cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn")),
lab = fct_recode(
outcome,
MMSE = "mmse_total",
CDR = "cdr",
"Executive\n Function" = "ef",
"Verbal\n Ability" = "va",
Memory = "em",
"Cortical \nThickness" = "meta_roi",
"Hippo. \nVolume" = "z_hippcampul_vol",
"Aβ40" = "ab40",
"Aβ42" = "ab42",
"Aβ42/Aβ40" = "ab42_ab40",
"Total Tau" = "tau",
"pTau" = "ptau",
NfL = "nfl",
Healthy = "cdx_cn_vs_ci",
MCI = "cdx_mci_vs_cn",
Dementia = "cdx_dem_vs_cn",
"Cog. Impair." = "cdx_ci_vs_cn"
),
lab = fct_inorder(lab))
```
### Z-test
```{r}
compare_coefs = function(b1, b2, se1, se2, return_pval = FALSE) {
z = (b1 - b2) / sqrt(se1^2 + se2^2)
if (return_pval) {
return(2 * (1 - pnorm(abs(z))))
} else {
return(z)
}
}
res_diff = res_crs %>%
filter(Parameter %in% c("mcaide", "cogd", "libra", "whicap")) %>%
select(term, lab, pop, Std_Coefficient, std.error, category) %>%
filter(!lab %in% c("Aβ42", "Aβ40")) %>%
group_by(term, lab, category) %>%
group_modify(~ {
pops = unique(.x$pop[.x$pop != "all"])
if(length(pops) < 2) return(tibble())
pairs = t(combn(pops, 2)) %>% as.data.frame()
colnames(pairs) = c("pop1", "pop2")
map_dfr(1:nrow(pairs), function(i) {
p1 = pairs$pop1[i]
p2 = pairs$pop2[i]
b1 = .x$Std_Coefficient[.x$pop == p1]
b2 = .x$Std_Coefficient[.x$pop == p2]
se1 = .x$std.error[.x$pop == p1]
se2 = .x$std.error[.x$pop == p2]
z = compare_coefs(b1, b2, se1, se2)
p = compare_coefs(b1, b2, se1, se2, return_pval = TRUE)
tibble(
pop1 = p1,
pop2 = p2,
z_score = z,
p_diff = p
)
})
}) %>%
ungroup() %>%
mutate(p_diff_fdr = p.adjust(p_diff, method = "fdr")) %>%
mutate(
term = recode(term, mcaide = "mCAIDE", whicap = "WHICAP", libra = "LIBRA", cogd = "CogDrisk"),
term = factor(term, levels = c("mCAIDE", "WHICAP", "LIBRA", "CogDrisk")),
pop1 = recode(pop1, "Black" = "AA", "Hispanic" = "LA", "NHW" = "NHW"),
pop2 = recode(pop2, "Black" = "AA", "Hispanic" = "LA", "NHW" = "NHW"),
pop1 = factor(pop1, levels = c("NHW", "LA", "AA")),
pop2 = factor(pop2, levels = c("NHW", "LA", "AA"))) %>%
droplevels()
res_diff = res_diff %>%
mutate(label = case_when(p_diff_fdr < 0.001 ~ "***", p_diff_fdr < 0.01 ~ "**", p_diff_fdr < 0.05 ~ "*")) %>%
rename(
Model = term,
Outcome = lab,
Category = category,
Pop1 = pop1,
Pop2 = pop2,
`Z-score` = z_score,
`P-Value` = p_diff,
`P-Value FDR. adj.` = p_diff_fdr,
Significance = label
)
res_sig = res_diff %>%
filter(p_diff_fdr < 0.05, (pop1 == "NHW" | pop2 == "NHW"))
res_diff_all = res_all %>%
select(term, outcome, pop, Std_Coefficient, std.error) %>%
filter(outcome != "cdx_cn_vs_ci") %>%
group_by(term, outcome) %>%
group_modify(~ {
pops = unique(.x$pop)
pairs = t(combn(pops, 2)) %>% as.data.frame()
colnames(pairs) = c("pop1", "pop2")
map_dfr(1:nrow(pairs), function(i) {
p1 = pairs$pop1[i]
p2 = pairs$pop2[i]
b1 = .x$Std_Coefficient[.x$pop == p1]
b2 = .x$Std_Coefficient[.x$pop == p2]
se1 = .x$std.error[.x$pop == p1]
se2 = .x$std.error[.x$pop == p2]
z = compare_coefs(b1, b2, se1, se2)
p = compare_coefs(b1, b2, se1, se2, return_pval = TRUE)
tibble(
pop1 = p1,
pop2 = p2,
z_score = z,
p_diff = p)
})
}) %>%
ungroup() %>%
mutate(p_diff_fdr = p.adjust(p_diff, method = "fdr"))
```
```{r}
# publication table - main + sensitivity analysis
coef_table_pred = bind_rows(res_crs, res_crs_sva) %>%
select(pop, term, category, lab, p.value, p.fdr, Parameter, Std_Coefficient, CI_low, CI_high) %>%
filter(Parameter %in% c("cogd", "libra", "whicap", "mcaide", "cogd_sva", "whicap_sva", "mcaide_sva")) %>%
filter(!lab %in% c("Healthy", "Aβ42", "Aβ40")) %>%
mutate(term = recode(term,
"cogd" = "CogDRisk",
"libra" = "LIBRA",
"whicap" = "WHICAP",
"mcaide" = "mCAIDE",
"cogd_sva" = "CogDRisk (sensitivity)",
"libra_sva" = "LIBRA (sensitivity)",
"whicap_sva" = "WHICAP (sensitivity)",
"mcaide_sva" = "mCAIDE (sensitivity)"),
lab = fct_recode(lab,
"Executive Function" = "Executive\n Function",
"Verbal Ability" = "Verbal\n Ability",
"Cortical Thickness" = "Cortical \nThickness",
"Hippocampal Volume" = "Hippo. \nVolume")
) %>%
rename(
Race = pop,
CRS = term,
Category = category,
Outcome = lab,
`P-value` = p.value,
`P-value, FDR adj.` = p.fdr,
β = Std_Coefficient,
`Lower Conf. Interval` = CI_low,
`Upper Conf. Interval` = CI_high
) %>%
select(-Parameter)
```
### OR
```{r}
# diagnosis only
color = c(AA = "#E41A1C", LA = "#377EB8", NHW = "#984EA3", all = "black")
diagnosis = ggplot(res_crs %>%
filter(
pop == "all",
outcome %in% c("cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn"),
term %in% c("mcaide", "whicap", "libra", "cogd")) %>%
mutate(
outcome = recode(outcome,
cdx_mci_vs_cn = "MCI",
cdx_dem_vs_cn = "Dementia",
cdx_ci_vs_cn = "MCI + Dementia"),
term = recode(term,
mcaide = "mCAIDE",
whicap = "WHICAP",
libra = "LIBRA",
cogd = "CogDrisk"),
outcome = factor(outcome, levels = c("MCI", "Dementia", "MCI + Dementia")),
term = factor(term, levels = c("mCAIDE", "WHICAP", "LIBRA", "CogDrisk")),
sig_fdr = factor(sig_fdr, levels = c(TRUE, FALSE))
),
aes(x = outcome, y = or, color = term, group = term, alpha = sig_fdr)) +
geom_point(position = position_dodge(width = 0.5), size = 1.5) +
geom_errorbar(aes(ymin = lci, ymax = uci),
position = position_dodge(width = 0.5),
width = 0, size = 0.8) +
geom_hline(yintercept = 1, linetype = 2) +
scale_alpha_manual(values = c(`TRUE` = 1, `FALSE` = 0.1)) +
scale_color_manual(values = c("mCAIDE" = "#003DE6",
"WHICAP" = "#d95f02",
"LIBRA" = "#F22424",
"CogDrisk" = "#005713")) +
ggtitle("All") +
labs(x = "", y = "Odds Ratio", color = "CRS", alpha = "FDR p < 0.05") +
theme_bw() +
theme(
axis.text.x = element_text(size = 8, angle = 0, hjust = 0.5),
axis.text.y = element_text(size = 8),
axis.title = element_text(size = 8),
plot.title = element_text(size = 10, face = "bold", hjust = 0.5),
legend.title = element_text(size = 8),
legend.position = "none",
strip.background = element_blank(),
strip.text = element_text(face = "bold", size = 10)
)
diagnosis_race = ggplot(res_crs %>%
filter(outcome %in% c("cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn"),
term %in% c("mcaide", "whicap", "libra", "cogd")) %>%
mutate(
outcome = recode(outcome,
cdx_mci_vs_cn = "MCI",
cdx_dem_vs_cn = "Dementia",
cdx_ci_vs_cn = "MCI + Dementia"),
outcome = factor(outcome, levels = c("MCI", "Dementia", "MCI + Dementia")),
term = recode(term,
mcaide = "mCAIDE",
whicap = "WHICAP",
libra = "LIBRA",
cogd = "CogDrisk"),
term = factor(term, levels = c("mCAIDE", "WHICAP", "LIBRA", "CogDrisk")),
pop = recode(pop, "Black" = "AA", "Hispanic" = "LA", "NHW" = "NHW", "all" = "all"),
pop = factor(pop, levels = c("NHW", "LA", "AA", "all")),
sig_fdr = factor(sig_fdr, levels = c(TRUE, FALSE))
),
aes(x = outcome, y = or, color = pop, alpha = sig_fdr)) +
geom_point(aes(group = interaction(term, pop)),
position = position_dodge(width = 0.6), size = 1.5) +
geom_errorbar(aes(ymin = lci, ymax = uci, group = interaction(term, pop)),
position = position_dodge(width = 0.6), width = 0, size = 0.8) +
geom_hline(yintercept = 1, linetype = 2) +
scale_alpha_manual(values = c(`TRUE` = 1, `FALSE` = 0.1)) +
scale_color_manual(values = color) +
labs(x = "", y = "OR (95% CI)", color = "Population", alpha = "FDR p < 0.05") +
facet_wrap(~term, nrow = 1) +
theme_bw() +
theme(
axis.text.x = element_text(size = 6, angle = 0, hjust = 0.5),
axis.text.y = element_text(size = 8),
axis.title = element_text(size = 8),
plot.title = element_text(size = 8, hjust = 0.5),
legend.position = "bottom",
strip.background = element_blank(),
strip.text = element_text(face = "bold", size = 10)
)
plot_all = res_crs %>%
filter(Parameter %in% c("mcaide", "whicap", "libra", "cogd"),
category %in% c("Neuroimaging", "Plasma \nBiomarkers", "Cognitive \nFunction")) %>%
filter(lab != "Aβ42", lab != "Aβ40") %>%
mutate(
pop = recode(pop, "Black" = "AA", "Hispanic" = "LA", "NHW" = "NHW", "all" = "all"),
pop = factor(pop, levels = c("NHW", "LA", "AA", "all")),
sig_fdr = factor(sig_fdr, levels = c(TRUE, FALSE)),
Parameter = recode(Parameter,
"mcaide" = "mCAIDE",
"whicap" = "WHICAP",
"libra" = "LIBRA",
"cogd" = "CogDRisk"),
Parameter = factor(Parameter, levels = c(
"mCAIDE", "WHICAP", "LIBRA", "CogDRisk")),
category = recode(category,
"Neuroimaging" = "Neuroimaging",
"Plasma \nBiomarkers" = "Plasma Biomarkers",
"Cognitive \nFunction" = "Cognitive Function"),
lab = recode(lab, "Hippo. \nVolume" = "Hippocampal \nVolume",
"WMH" = "White Matter \nHyperintensity"),
lab = factor(lab, levels = c(
"MMSE", "CDR", "Memory", "Verbal\n Ability", "Executive\n Function",
"Hippocampal \nVolume", "Cortical \nThickness",
"NfL", "Total Tau", "pTau", "Aβ42/Aβ40"))) %>%
ggplot(aes(x = lab, y = Std_Coefficient, color = pop, alpha = sig_fdr)) +
geom_hline(yintercept = 0, linetype = 2) +
geom_point(position = position_dodge(width = 0.6), size = 1.5) +
geom_errorbar(aes(ymin = CI_low, ymax = CI_high),
position = position_dodge(width = 0.6), width = 0, size = 0.8) +
facet_grid(Parameter ~ category, scales = "free", space = "free_x") +
scale_color_manual(values = color) +
scale_alpha_manual(values = c(`TRUE` = 1, `FALSE` = 0.3)) +
#scale_y_continuous(limits = c(-0.52, 0.48), breaks = seq(-0.5, 0.5, 0.25)) +
labs(
x = "",
y = "Beta (95% CI)",
color = "Population",
alpha = "FDR p < 0.05"
) +
theme_bw() +
theme(
axis.text.x = element_text(size = 8, angle = 45, hjust = 1),
axis.text.y = element_text(size = 8),
axis.title = element_text(size = 8),
strip.background = element_blank(),
strip.text = element_text(face = "bold", size = 10),
legend.text = element_text(size = 10),
legend.title = element_text(size = 10),
legend.position = "bottom")
```
### AUC + R2
```{r}
adjust_crs = function(df_models, pop_label = NULL) {
df_data = if (!is.null(pop_label)) data %>% filter(pop == pop_label) else data
df_models %>%
mutate(
pop = ifelse(is.null(pop_label), "ALL", pop_label),
eq_crs_only = case_when(
model %in% c("base", "base_noapoe") ~ eq,
TRUE ~ str_replace(eq, "apoe \\+ ", "")
),
res_crs_only = map(eq_crs_only, ~ glm(.x, data = df_data, family = binomial)),
auc_crs_only = map_dbl(res_crs_only, ~ {
y = model.response(model.frame(.x))
p = predict(.x, type = "response")
if (length(unique(y)) > 1) pROC::auc(pROC::roc(y, p)) else NA_real_
}),
nagelkerker2_crs_only = map_dbl(res_crs_only, ~ {
if (length(unique(model.response(model.frame(.x)))) > 1)
pscl::pR2(.x)[["McFadden"]]
else NA_real_
})
)
}
afr_cdx_prs2 = adjust_crs(afr_cdx_prs, pop_label = "AFR") %>% mutate(pop = "Black")
amr_cdx_prs2 = adjust_crs(amr_cdx_prs, pop_label = "AMR") %>% mutate(pop = "Hispanic")
eur_cdx_prs2 = adjust_crs(eur_cdx_prs, pop_label = "EUR") %>% mutate(pop = "NHW")
all_cdx_prs2 = adjust_crs(all_cdx_prs) %>% mutate(pop = "all")
all_comb_cdx2 = bind_rows(all_cdx_prs2, afr_cdx_prs2, amr_cdx_prs2, eur_cdx_prs2)
plot_roc_outcome = function(df, outcome_name, crs_type = "CRS", pop_filter = NULL) {
crs_colors = c(
"Demographics+APOE" = "#7A3900",
"Demographics" = "#7A1100",
"mCAIDE" = "#003DE6",
"mCAIDE-" = "#003DE6",
"WHICAP" = "#d95f02",
"WHICAP-" = "#d95f02",
"LIBRA" = "#F22424",
"LIBRA-" = "#F22424",
"CogDRisk" = "#005713",
"CogDRisk-" = "#005713"
)
df_out = df %>% filter(outcome == outcome_name)
if (!is.null(pop_filter)) df_out = df_out %>% filter(pop == pop_filter)
df_out = df_out %>%
mutate(
outcome = recode(outcome,
"cdx_mci_vs_cn" = "MCI",
"cdx_dem_vs_cn" = "Dementia",
"cdx_ci_vs_cn" = "MCI + Dementia"),
model = recode(model,
"base" = "Demographics+APOE",
"base_noapoe" = "Demographics",
"libra" = "LIBRA",
"mcaide" = "mCAIDE",
"whicap" = "WHICAP",
"cogd" = "CogDRisk",
"libra_sva" = "LIBRA-",
"mcaide_sva" = "mCAIDE-",
"whicap_sva" = "WHICAP-",
"cogd_sva" = "CogDRisk-")) %>%
mutate(model = factor(model, levels = c(
"CogDRisk", "LIBRA", "WHICAP", "mCAIDE", "Demographics", "Demographics+APOE",
"mCAIDE-", "WHICAP-", "LIBRA-", "CogDRisk-")))
if (crs_type == "CRS") {
df_out = df_out %>% filter(!grepl("-$", model) | model %in% c("Demographics+APOE", "Demographics"))
} else if (crs_type == "CRS-") {
df_out = df_out %>% filter(grepl("-$", model) | model %in% c("Demographics+APOE", "Demographics"))
}
desired_order = c(
"CogDRisk", "LIBRA", "WHICAP", "mCAIDE", "Demographics", "Demographics+APOE")
df_out$model = factor(df_out$model, levels = desired_order)
# compute ROC curves
df_out = df_out %>%
mutate(
roc_obj = map(res_crs_only, ~ {
y = model.response(model.frame(.x))
p = predict(.x, type = "response")
if (length(unique(y)) > 1) pROC::roc(y, p) else NULL
}),
auc_val = map_dbl(roc_obj, ~ if (!is.null(.x)) as.numeric(pROC::auc(.x)) else NA_real_),
auc_label = paste0(model, ": AUC = ", round(auc_val, 2)),
roc_df = map(roc_obj, ~ if (!is.null(.x)) tibble(fpr = 1 - .x$specificities,
tpr = .x$sensitivities) else NULL)
)
roc_plot_df = df_out %>%
select(model, auc_val, auc_label, roc_df) %>%
tidyr::unnest(roc_df)
format_r2 = function(x) {
ifelse(x < 0.01, formatC(x, format = "f", digits = 4), formatC(x, format = "f", digits = 3))
}
auc_labels_df = df_out %>%
select(model, auc_val, nagelkerker2_crs_only) %>%
distinct(model, auc_val, nagelkerker2_crs_only) %>%
arrange(factor(model, levels = desired_order)) %>%
mutate(
auc_label = paste0(model, ": ", round(auc_val, 2),
"; ", format_r2(nagelkerker2_crs_only)),
fpr = 0.98, tpr = 0.02 + 0.05 * (row_number())
)
ggplot(roc_plot_df, aes(x = fpr, y = tpr, color = model)) +
geom_line(size = 0.5) +
geom_abline(intercept = 0, slope = 1, linetype = "dashed", color = "gray50") +
geom_text(
data = auc_labels_df,
aes(x = fpr, y = tpr, label = auc_label),
inherit.aes = FALSE, hjust = 1, vjust = 0, size = 2
) +
scale_color_manual(values = crs_colors[levels(df_out$model)]) +
labs(x = "False Positive Rate",
y = "True Positive Rate",
color = NULL,
title = unique(df_out$outcome)) +
theme_bw() +
theme(
axis.title = element_text(size = 6),
legend.position = "none",
plot.title = element_text(hjust = 0.5, size = 8, face = "bold")
)
}
#MCI
roc_mci_all = plot_roc_outcome(all_comb_cdx2, "cdx_mci_vs_cn", crs_type = "CRS", pop_filter = "all")
roc_mci_aa = plot_roc_outcome(all_comb_cdx2, "cdx_mci_vs_cn", crs_type = "CRS", pop_filter = "Black")
roc_mci_la = plot_roc_outcome(all_comb_cdx2, "cdx_mci_vs_cn", crs_type = "CRS", pop_filter = "Hispanic")
roc_mci_nhw = plot_roc_outcome(all_comb_cdx2, "cdx_mci_vs_cn", crs_type = "CRS", pop_filter = "NHW")
#Dementia
roc_dem_all = plot_roc_outcome(all_comb_cdx2, "cdx_dem_vs_cn", crs_type = "CRS", pop_filter = "all")
roc_dem_aa = plot_roc_outcome(all_comb_cdx2, "cdx_dem_vs_cn", crs_type = "CRS", pop_filter = "Black")
roc_dem_la = plot_roc_outcome(all_comb_cdx2, "cdx_dem_vs_cn", crs_type = "CRS", pop_filter = "Hispanic")
roc_dem_nhw = plot_roc_outcome(all_comb_cdx2, "cdx_dem_vs_cn", crs_type = "CRS", pop_filter = "NHW")
#MCI+Dementia
roc_ci_all = plot_roc_outcome(all_comb_cdx2, "cdx_ci_vs_cn", crs_type = "CRS", pop_filter = "all")
roc_ci_aa = plot_roc_outcome(all_comb_cdx2, "cdx_ci_vs_cn", crs_type = "CRS", pop_filter = "Black")
roc_ci_la = plot_roc_outcome(all_comb_cdx2, "cdx_ci_vs_cn", crs_type = "CRS", pop_filter = "Hispanic")
roc_ci_nhw = plot_roc_outcome(all_comb_cdx2, "cdx_ci_vs_cn", crs_type = "CRS", pop_filter = "NHW")
shared_legend = get_legend(
roc_mci_all +
theme(
legend.position = "bottom",
legend.title = element_text(size = 8, face = "bold"),
legend.text = element_text(size = 6)) +
guides(color = guide_legend(nrow = 1)))
pageCreate(width = 9, height = 12.5, default.units = "inches")
plotGG(
plot = roc_mci_all,
x = 0, y = 0, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotText(
label = "A)", x = 0.1, y = 0.1,
just = c("left", "top"), fontface = "bold",
fontsize = 12, default.units = "inches"
)
plotGG(
plot = roc_dem_all,
x = 3, y = 0, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_ci_all,
x = 6, y = 0, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_mci_aa,
x = 0, y = 3, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotText(
label = "B)", x = 0.1, y = 3.1,
just = c("left", "top"), fontface = "bold",
fontsize = 12, default.units = "inches"
)
plotGG(
plot = roc_dem_aa,
x = 3, y = 3, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_ci_aa,
x = 6, y = 3, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_mci_la,
x = 0, y = 6, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotText(
label = "C)", x = 0.1, y = 6.1,
just = c("left", "top"), fontface = "bold",
fontsize = 12, default.units = "inches"
)
plotGG(
plot = roc_dem_la,
x = 3, y = 6, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_ci_la,
x = 6, y = 6, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_mci_nhw,
x = 0, y = 9, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotText(
label = "D)", x = 0.1, y = 9.1,
just = c("left", "top"), fontface = "bold",
fontsize = 12, default.units = "inches"
)
plotGG(
plot = roc_dem_nhw,
x = 3, y = 9, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
plot = roc_ci_nhw,
x = 6, y = 9, width = 3, height = 3,
just = c("left", "top"), default.units = "inches"
)
plotGG(
shared_legend,
x = 2, y = 12, width = 6, height = 0.5,
just = c("left", "top")
)
pageGuideHide()
```
```{r}
# publication table for AUC & R2 for linear and logistic regression
adjust_crs_linear = function(df_models, pop_label = NULL) {
df_data = if (!is.null(pop_label)) data %>% filter(pop == pop_label) else data
df_models %>%
mutate(
pop = ifelse(is.null(pop_label), "ALL", pop_label),
eq_crs_only = case_when(
model %in% c("base", "base_noapoe") ~ eq,
TRUE ~ str_replace(eq, "apoe \\+ ", "")
),
res_crs_only = map(eq_crs_only, ~ lm(as.formula(.x), data = df_data)),
r2_crs_only = map_dbl(res_crs_only, ~ {
summ <- summary(.x)
if (!is.null(summ$r.squared)) summ$r.squared else NA_real_
})
)
}
afr_num_prs2 = adjust_crs_linear(afr_num_prs, pop_label = "AFR") %>% mutate(pop = "Black")
amr_num_prs2 = adjust_crs_linear(amr_num_prs, pop_label = "AMR") %>% mutate(pop = "Hispanic")
eur_num_prs2 = adjust_crs_linear(eur_num_prs, pop_label = "EUR") %>% mutate(pop = "NHW")
all_num_prs2 = adjust_crs_linear(all_num_prs) %>% mutate(pop = "all")
all_comb_num2 = bind_rows(all_num_prs2, afr_num_prs2, amr_num_prs2, eur_num_prs2)
res_pfm = bind_rows(
select(all_comb_cdx2, model, pop, outcome, auc_crs_only, nagelkerker2_crs_only),
select(all_comb_num2, model, pop, outcome, r2_crs_only)) %>%
filter(model %in% c("base", "base_noapoe", "cogd", "libra", "whicap", "mcaide", "cogd_sva", "libra_sva", "whicap_sva", "mcaide_sva")) %>%
filter(!outcome %in% c("cdx_hypertension", "cdx_dyslipidemia", "cdx_diabetes", "bmi", "cdx_anxiety", "cdx_depression", "cdx_cvd", "wmh_volume_log",
"ab40", "ab42", "cdx_cn_vs_ci")) %>%
arrange(pop, model) %>%
mutate(
model = recode(model,
"base_noapoe" = "Demographics",
"base" = "Demographics + APOE",
"cogd" = "CogDRisk",
"libra" = "LIBRA",
"whicap" = "WHICAP",
"mcaide" = "mCAIDE",
"cogd_sva" = "CogDRisk (sensitivity)",
"libra_sva" = "LIBRA (sensitivity)",
"whicap_sva" = "WHICAP (sensitivity)",
"mcaide_sva" = "mCAIDE (sensitivity)"),
category = case_when(
outcome %in% c("va", "em", "ef", "mmse_total", "cdr") ~ "Cognitive \nFunction",
outcome %in% c("tau", "ptau", "nfl", "ab42_ab40") ~ "Plasma \nBiomarkers",
outcome %in% c("meta_roi", "z_hippcampul_vol") ~ "Neuroimaging",
outcome %in% c("cdx_mci_vs_cn", "cdx_dem_vs_cn", "cdx_ci_vs_cn") ~ "Medical \nHistory"),
lab = fct_recode(
outcome,
MMSE = "mmse_total",
CDR = "cdr",
"Executive\n Function" = "ef",
"Verbal\n Ability" = "va",
Memory = "em",
"Cortical \nThickness" = "meta_roi",
"Hippo. \nVolume" = "z_hippcampul_vol",
"Aβ42/Aβ40" = "ab42_ab40",
"Total Tau" = "tau",
"pTau" = "ptau",
NfL = "nfl",
MCI = "cdx_mci_vs_cn",
Dementia = "cdx_dem_vs_cn",
"Cog. Impair." = "cdx_ci_vs_cn"),
lab = fct_inorder(lab)
) %>%
rename(
Model = model,
Race = pop,
Outcome = lab,
Category = category,
AUC = auc_crs_only,
`Nagelkerke R2` = nagelkerker2_crs_only,
R2 = r2_crs_only
) %>%
select(-outcome) %>%
select(Model, Race, Outcome, Category, AUC, `Nagelkerke R2`, R2)
```
### Demographics
```{r}
# table 1
summary_df = data %>%
filter(!is.na(cdx_cog)) %>%
group_by(cdx_cog) %>%
summarise(
Total = n(),
Age = sprintf("%.1f (± %.1f)", mean(age, na.rm = TRUE), sd(age, na.rm = TRUE)),
Female = sprintf("%d (%.1f%%)", sum(sex == 1, na.rm = TRUE), 100 * mean(sex == 1, na.rm = TRUE)),
Male = sprintf("%d (%.1f%%)", sum(sex == 0, na.rm = TRUE), 100 * mean(sex == 0, na.rm = TRUE)),
`ε4 Carrier` = sprintf("%d (%.1f%%)", sum(apoe4 == 1, na.rm = TRUE), 100 * mean(apoe4 == 1, na.rm = TRUE)),
`ε4 Noncarrier` = sprintf("%d (%.1f%%)", sum(apoe4 == 0, na.rm = TRUE), 100 * mean(apoe4 == 0, na.rm = TRUE)),
mCAIDE = sprintf("%.3f (± %.2f)", mean(mcaide, na.rm = TRUE), sd(mcaide, na.rm = TRUE)),
WHICAP = sprintf("%.3f (± %.2f)", mean(whicap, na.rm = TRUE), sd(whicap, na.rm = TRUE)),
LIBRA = sprintf("%.3f (± %.2f)", mean(libra, na.rm = TRUE), sd(libra, na.rm = TRUE)),
CogDRisk = sprintf("%.3f (± %.2f)", mean(cogd, na.rm = TRUE), sd(cogd, na.rm = TRUE)),
MMSE = sprintf("%.2f (± %.2f)", mean(mmse_total, na.rm = TRUE), sd(mmse_total, na.rm = TRUE)),
CDR = sprintf("%.4f (± %.2f)", mean(cdr, na.rm = TRUE), sd(cdr, na.rm = TRUE)),
`Executive\n Function` = sprintf("%.2f (± %.2f)", mean(ef, na.rm = TRUE), sd(ef, na.rm = TRUE)),
`Verbal\n Ability` = sprintf("%.2f (± %.2f)", mean(va, na.rm = TRUE), sd(va, na.rm = TRUE)),
Memory = sprintf("%.2f (± %.2f)", mean(em, na.rm = TRUE), sd(em, na.rm = TRUE)),
`Cortical \nThickness` = sprintf("%.2f (± %.2f)", mean(meta_roi, na.rm = TRUE), sd(meta_roi, na.rm = TRUE)),
`Hippo. \nVolume` = sprintf("%.2f (± %.2f)", mean(z_hippcampul_vol, na.rm = TRUE), sd(z_hippcampul_vol, na.rm = TRUE)),
`Aβ42/Aβ40` = sprintf("%.3f (± %.2f)", mean(ab42_ab40, na.rm = TRUE), sd(ab42_ab40, na.rm = TRUE)),
`Total Tau` = sprintf("%.3f (± %.2f)", mean(tau, na.rm = TRUE), sd(tau, na.rm = TRUE)),
pTau = sprintf("%.2f (± %.2f)", mean(ptau, na.rm = TRUE), sd(ptau, na.rm = TRUE)),
NfL = sprintf("%.3f (± %.2f)", mean(nfl, na.rm = TRUE), sd(nfl, na.rm = TRUE))
) %>%
tibble::column_to_rownames("cdx_cog") %>%
t() %>%
as.data.frame() %>%
tibble::rownames_to_column("Variable")
# racial breakdown rows
race_summary = data %>%
filter(!is.na(cdx_cog), race %in% c("NHW", "Hispanic", "Black")) %>%
group_by(cdx_cog, race) %>%
summarise(n = n(), .groups = "drop_last") %>%
mutate(
perc = round(100 * n / sum(n), 1),
value = sprintf("%d (%.1f%%)", n, perc)
) %>%
select(cdx_cog, Variable = race, value) %>%
tidyr::pivot_wider(names_from = cdx_cog, values_from = value)
summary_df = bind_rows(summary_df, race_summary)
summary_df = summary_df %>%
mutate(
Group = case_when(
Variable %in% c("Total", "Case", "Control") ~ "AD",
Variable == "Age" ~ "Age",
Variable %in% c("Female", "Male") ~ "Sex",
Variable %in% c("NHW", "Hispanic", "Black") ~ "Race",
Variable %in% c("ε4 Carrier", "ε4 Noncarrier") ~ "APOE",
Variable %in% c("mCAIDE", "WHICAP", "LIBRA", "CogDRisk") ~ "CRS",
Variable %in% c("MMSE", "CDR", "Executive\n Function", "Verbal\n Ability", "Memory") ~ "Cognition",
Variable %in% c("Cortical \nThickness", "Hippo. \nVolume") ~ "Imaging",
Variable %in% c("Aβ40", "Aβ42", "Aβ42/Aβ40", "Total Tau", "pTau", "NfL") ~ "Biomarker",
TRUE ~ NA_character_
)) %>%
rename(
Characteristics = Variable,
`Cognitively Normal` = "0",
`Mild Cognitive Impairment` = "1",
Dementia = "2"
) %>%
select(-Group)
```

crs_analysis.qmd at commit aea67ca, under MIT · at the source

Overview

Authors: Meri Okorie^1,2, Xiaqing Jiang², Kristine Yaffe^1,2,3,4,5, Jennifer S Yokoyama^1,3,6, Shea J Andrews^1,2, for the Health and Aging Brain Study–Health Disparities

ORCID iDs: Meri Okorie, Jennifer S Yokoyama, Shea J Andrews

Edward and Pearl Fein Memory and Aging Center, University of California, San Francisco, San Francisco, California, USA
Department of Psychiatry and Behavioural Sciences, University of California, San Francisco, San Francisco, California, USA
Department of Neurology and Weill Institute for Neurosciences, University of California, San Francisco, San Francisco, California, USA
Department of Epidemiology and Biostatistics, University of California, San Francisco, San Francisco, California, USA
San Francisco Veterans Affairs Health Care System, San Francisco, California, USA
Department of Radiology and Biomedical Imaging, University of California, San Francisco, San Francisco, California, USA

Institutions: University of California, San Francisco (United States); San Francisco VA Health Care System (United States)

Journal: Alzheimer's & dementia : the journal of the Alzheimer's Association, volume 22, issue 8, article e71567

Dates: received 4 February 2026; accepted 4 May 2026; published online 28 July 2026; in print August 2026

Type: Research article · Language: English

License: CC BY-NC-ND

Identifiers: DOI 10.1002/alz.71567 · PMID 42522069 · PMCID PMC13415752 · OpenAlex W7171670744

Open access: hybrid, a free copy (OpenAlex)

Status: code verified

Categories: PET / SPECT (modality), human (organism), Alzheimer's / dementia (population), clinical / translational (subfield)

Methods: Statistics, Machine learning, Physiology & signal measures, Preprocessing

Keywords: Alzheimer's disease, biomarkers, clinical risk score, health disparities, modifiable risk factors

MeSH: Alzheimer Disease*, Dementia*, Endophenotypes*, Aged, Aged, 80 and over, Amyloid beta-Peptides, Female, Humans, Male, Peptide Fragments, Positron-Emission Tomography, Risk Factors, tau Proteins (* major topic)

Topic: Dementia and Cognitive Impairment Research (Psychiatry and Mental health, Medicine), according to OpenAlex

Funding: NIA NIH HHS (R01 AG057234, U24 AG072122, P30AG062422, U19 AG078109, R01 AG062588, U19 AG079774, U19AG079774, R01 AG054073, R01 AG058533, P01 AG019724, P01AG019724, R01AG057234, P30 AG062422, R01 AG070862); Alzheimer's Association; Genentech; NIBIB NIH HHS (P41 EB015922); Global Brain Health Institute; National Institute on Aging (R01AG057234, P30AG062422, U19AG079774, P01AG019724); Alzheimer's Association; National Alzheimer's Coordinating Center (U24AG072122); National Alzheimer’s Coordinating Center (U24AG072122); NINDS NIH HHS (U54 NS123985, U54NS123985); French Foundation; National Institute of Neurological Disorders and Stroke (U54NS123985); Rainwater Charitable Foundation; Mary Oakley Foundation

Citations: not cited yet (Europe PMC); 56 references in the paper

Abstract

The abstract is not reproduced here: the paper's license (CC BY-NC-ND) does not allow it. Read it in the paper, at the publisher or on Europe PMC.

Repository

Its files are read in the Code ↔ Paper reader above, with 13 matches between paragraphs and lines of code.

AndrewsLabUCSF/CRS-analysis

License: MIT

State: the link answers, verified on 26 September 2026

Evidence: files inventoried

Commit: aea67caca4b114ff2aa32bb4bf5d244127574e7b, 25 March 2026

Languages: Quarto (4)

Size: 11 files, 4 scripts

Software Heritage: not archived

Found in: “CODE AVAILABILITY”

Holds: README, license file, documentation, 4 notebooks

Not found: CITATION.cff, environment file, tests, continuous integration

Tools: tidyverse (4 files), broom (3 files), ggplot2 (2 files), pROC (2 files), cowplot (1 file), data.table (1 file), easystats (1 file), patchwork (1 file), randomForest (1 file)

Availability: 1 check, the latest on 26 September 2026: the link answers

26 September 2026: the link answers

6 files

workflow/scripts/crs_analysis.qmd, Quarto, 2,407 lines, 7 matches
workflow/scripts/crs_standardization.qmd, Quarto, 471 lines, 3 matches
workflow/scripts/cutoff_analysis.qmd, Quarto, 401 lines, 1 match
workflow/scripts/imputation.qmd, Quarto, 472 lines, 2 matches
LICENSE, License, 21 lines
README.md, Text, 154 lines

At the source: github.com/AndrewsLabUCSF/CRS-analysis

Code availability statement

The paper has a code availability statement. Its license (CC BY-NC-ND) does not allow reproducing it here; in short, from what the harvester recognized in it:

it points to the authors' code: AndrewsLabUCSF/CRS-analysis

Read it in the paper: doi.org/10.1002/alz.71567.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
4 scripts, each with its path and the digest of its content;
13 matches between paragraphs of the paper and lines of the code (method lexical-v1);
neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 6 authors, 5 keywords, 13 MeSH terms, 14 funders, 51 references.

Cite

This paper

Okorie, M., Jiang, X., Yaffe, K., Yokoyama, J. S., Andrews, S. J., & for the Health and Aging Brain Study–Health Disparities. (2026). Associations of dementia polyexposure scores to Alzheimer's disease endophenotypes in a diverse population. Alzheimer's & dementia : the journal of the Alzheimer's Association, 22(8), e71567. https://doi.org/10.1002/alz.71567

BibTeX

@article{okorie2026associations, author = {Okorie, Meri and Jiang, Xiaqing and Yaffe, Kristine and Yokoyama, Jennifer S and Andrews, Shea J and {for the Health and Aging Brain Study–Health Disparities}}, title = {{Associations of dementia polyexposure scores to Alzheimer's disease endophenotypes in a diverse population}}, journal = {Alzheimer's \& dementia : the journal of the Alzheimer's Association}, year = {2026}, month = aug, volume = {22}, number = {8}, pages = {e71567}, publisher = {Wiley}, issn = {1552-5260}, doi = {10.1002/alz.71567}, url = {https://doi.org/10.1002/alz.71567}, pmid = {42522069}, pmcid = {PMC13415752} }

RIS

TY - JOUR AU - Okorie, Meri AU - Jiang, Xiaqing AU - Yaffe, Kristine AU - Yokoyama, Jennifer S AU - Andrews, Shea J AU - for the Health and Aging Brain Study–Health Disparities TI - Associations of dementia polyexposure scores to Alzheimer's disease endophenotypes in a diverse population T2 - Alzheimer's & dementia : the journal of the Alzheimer's Association J2 - Alzheimers Dement PY - 2026 DA - 2026/08/01 VL - 22 IS - 8 SP - e71567 SN - 1552-5260 PB - Wiley DO - 10.1002/alz.71567 UR - https://doi.org/10.1002/alz.71567 LA - en ER -

CSL-JSON

{ "id": "10.1002/alz.71567", "type": "article-journal", "title": "Associations of dementia polyexposure scores to Alzheimer's disease endophenotypes in a diverse population", "container-title": "Alzheimer's & dementia : the journal of the Alzheimer's Association", "author": [ { "family": "Okorie", "given": "Meri" }, { "family": "Jiang", "given": "Xiaqing" }, { "family": "Yaffe", "given": "Kristine" }, { "family": "Yokoyama", "given": "Jennifer S" }, { "family": "Andrews", "given": "Shea J" }, { "literal": "for the Health and Aging Brain Study–Health Disparities" } ], "container-title-short": "Alzheimers Dement", "volume": "22", "issue": "8", "page": "e71567", "DOI": "10.1002/alz.71567", "PMID": "42522069", "PMCID": "PMC13415752", "ISSN": "1552-5260", "publisher": "Wiley", "URL": "https://doi.org/10.1002/alz.71567", "language": "en", "issued": { "date-parts": [ [ 2026, 8, 1 ] ] } }

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.