{card} collects datasets and tools for cardiovascular research data:
FDA MAUDE device adverse event reports, echocardiogram report text,
procedural billing codes, and clinical genetics annotations, with
cosinor rhythm models that are moving to a package of their own.
You can install the released version of {card} from
CRAN with:
install.packages("card")
And the development version from GitHub with:
# install.packages("devtools")
# devtools::install_github("shah-in-boots/card")
install.packages("card", repos = c('https://shah-in-boots.r-universe.dev', 'https://cloud.r-project.org'))
The intent for card is to assist in the analysis of cardiovascular and
epidemiological data, with a focus on cardiac-specific datasets and data
types.
library(card)
# ECG data
data("twins")
head(twins)
#> # A tibble: 6 × 23
#> # Groups: patid, hour [6]
#> patid age bmi race smoking hptn dm chf prevchd med_beta_blockers
#> <dbl> <dbl> <dbl> <fct> <fct> <fct> <fct> <fct> <fct> <fct>
#> 1 1 49 27.4 0 1 1 0 0 0 0
#> 2 1 49 27.4 0 1 1 0 0 0 0
#> 3 1 49 27.4 0 1 1 0 0 0 0
#> 4 1 49 27.4 0 1 1 0 0 0 0
#> 5 1 49 27.4 0 1 1 0 0 0 0
#> 6 1 49 27.4 0 1 1 0 0 0 0
#> # ℹ 13 more variables: med_antidepr <fct>, beck_total <dbl>, sad_bin <fct>,
#> # sad_cat <fct>, PETdiff_2 <fct>, dyxtime <dttm>, date <date>, hour <dbl>,
#> # rDYX <dbl>, sDYX <dbl>, HR <dbl>, CP <dbl>, zip <chr>
# Outcomes data
data("stress")
head(stress)
#> # A tibble: 6 × 11
#> id start stop head_ache_date_1 head_ache_date_2 head_ache_date_3
#> <dbl> <date> <date> <date> <date> <date>
#> 1 123 2014-08-26 2016-12-08 2016-12-08 NA NA
#> 2 117 2014-08-12 2016-10-30 2016-10-30 NA NA
#> 3 145 2014-10-22 2016-07-31 2016-06-01 2016-07-31 NA
#> 4 144 2014-10-21 2017-03-24 2015-02-12 NA NA
#> 5 204 2015-03-16 2015-04-03 NA NA NA
#> 6 283 2015-10-12 2016-01-19 NA NA NA
#> # ℹ 5 more variables: heart_ache_date_1 <date>, heart_ache_date_2 <date>,
#> # heart_ache_date_3 <date>, death <dbl>, broken_heart <dbl>
# VCG data
data("geh")
head(geh)
#> # A tibble: 6 × 70
#> pid hhp_id age sex age_cat systolic_bp_first systolic_bp_second
#> <dbl> <dbl> <dbl> <fct> <fct> <dbl> <dbl>
#> 1 200481 220946 42 0 0 162 165
#> 2 200489 224643 62 1 1 133 128
#> 3 200495 224845 84 0 2 100 99
#> 4 200052 222917 65 0 2 139 141
#> 5 200517 220638 64 1 1 166 163
#> 6 200623 224862 57 0 1 151 144
#> # ℹ 63 more variables: systolic_bp_third <dbl>, diastolic_bp_first <dbl>,
#> # diastolic_bp_second <dbl>, diastolic_bp_third <dbl>,
#> # pulse_rate_first <dbl>, pulse_rate_second <dbl>, height_cm <dbl>,
#> # weight_kg <dbl>, waist_cm <dbl>, dia_trt_allopdrug <hvn_lbll>,
#> # hbp_trt_allopdrug <hvn_lbll>, hyp_trt_allopdrug <hvn_lbll>,
#> # lab_hba1c <dbl>, lab_fasting_bg <dbl>, lab_fasting_insulin <dbl>,
#> # lab_tchol <dbl>, lab_ldlchol <dbl>, lab_hdlchol <dbl>, lab_triglyc <dbl>, …
{card} queries the FDA’s MAUDE database through openFDA, resolves the
free-text manufacturer and brand strings onto companies and ablation
platforms, and groups the coded problem terms into clinical complication
families. Every user-facing function starts with maude_.
events <- maude_query(device_brand_name = "FARAWAVE", limit = 500)
The manufacturer field names the plant rather than the company, and who owns a company changes with time, so the two are resolved separately:
maude_normalize_manufacturer(c("BIOSENSE WEBSTER, INC. (JUAREZ)", "FARAPULSE, INC."))
#> [1] "Biosense Webster" "Farapulse"
maude_resolve_owner("Farapulse", as_of = as.Date(c("2019-01-01", "2024-01-01")))
#> [1] "Farapulse" "Boston Scientific"
maude_term_to_complication(c("Cardiac Tamponade", "Pericardial Effusion"))
#> $pericardial
#> [1] "Cardiac Tamponade" "Pericardial Effusion"
maude_adjudicate() sends a report’s narrative to an {ellmer} chat
model and returns one set of 0/1 flags per candidate complication
family. vignette("maude") walks through the whole workflow.
cosinor() and its companions will leave {card} before version 1.0,
for a separate package for recurring and periodic events; see
?cosinor.
This modeling algorithm requires only base R, which allows internal
flexibility for modeling heuristics and improved efficiency. For the
user, it also allows standard modeling tools/approaches, and a flexible
user interface that accounts for individual/population analysis and
single/multiple component analysis.
There are other cosinor models available in R, however they do not
allow for simple multiple-component or individual/population analysis in
a method that is tidy (for both tidyverse and tidymodels
approaches).
m <- cosinor(rDYX ~ hour, twins, tau = c(24, 12))
summary(m)
#> Individual Cosinor Model
#> ------------------------------------------
#> Call:
#> cosinor(formula = rDYX ~ M + A1 * cos(2*pi*hour/24 + phi1) + A2 * cos(2*pi*hour/12 + phi2)
#>
#> Period(s): 24, 12
#>
#> Residuals:
#> Min. 1st Qu. Median Mean 3rd Qu. Max.
#> -3.00770 -0.52024 -0.03247 0.00000 0.48753 4.88552
#>
#> Coefficients:
#> Estimate Std. Error
#> mesor 2.8586510 0.006062639
#> amp1 0.2964114 0.008702368
#> phi1 -2.6542757 0.028468900
#> amp2 0.1302012 0.008542526
#> phi2 -3.6636921 0.065919951
#>
#> Zero-amplitude test: F = 353.599 on 4 and 16381 DF, p-value = < 2.2e-16
#> Design condition number: 1.5
ggcosinor(m)
#> This is a harmonic multiple-component cosinor object. The orthophase, bathyphase, and global amplitude were calculated.
#> Warning in regularize.values(x, y, ties, missing(ties)): collapsing to unique
#> 'x' values
#> Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
#> ℹ Please use `linewidth` instead.
#> ℹ The deprecated feature was likely used in the card package.
#> Please report the issue at <https://github.com/shah-in-boots/card/issues>.
#> This warning is displayed once per session.
#> Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
#> generated.
#> `geom_smooth()` using formula = 'y ~ s(x, bs = "cs")'
head(augment(m))
#> # A tibble: 6 × 8
#> y t x1 x2 z1 z2 .fitted .resid
#> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
#> 1 2.63 8 -0.5 -5.00e- 1 8.66e- 1 -8.66e- 1 3.22 -0.592
#> 2 2.42 9 -0.707 -1.84e-16 7.07e- 1 -1 e+ 0 3.21 -0.787
#> 3 1.81 10 -0.866 5 e- 1 5 e- 1 -8.66e- 1 3.15 -1.34
#> 4 2.01 11 -0.966 8.66e- 1 2.59e- 1 -5.00e- 1 3.08 -1.07
#> 5 1.63 12 -1 1 e+ 0 1.22e-16 -2.45e-16 3.01 -1.38
#> 6 1.95 13 -0.966 8.66e- 1 -2.59e- 1 5 e- 1 2.95 -0.996