In jcpsantiago/IScalc: Calculate Various Glucose Metabolism Indices

Below are some practical example showing how one can use the gmi function to calculate different insulin sensitivity indexes (ISI).

The functions included in this package expect data to be in vectors, which makes them to integrate with a tidyverse workflow. Below you will also find examples in case your data is not in a observation-per-row format, or what is known as tidy.

Loading the package

gmi will load like any standard R package:

library(gmi)

# dplyr will make it easy to apply the functions to whole datasets.
library(dplyr)
library(tidyr)

Calculating ISI with tidy data

If your data looks like the block below, continue reading this section, otherwise move on to the next section.

head(OGTT_long)

The data above has one observation per row, which is the ideal format to apply the gmi functions quickly.

As a first example I calculate the revised QUICKI for each subject in the included sample dataset OGTT_long. The ISI_fasting function will compute an index from fasting values; glu (glucose), ins (insulin) and index must be provided. Additionally, for the revised QUICKI index you will need to provide the nefa argument.

# OGTT_long %>%
#   filter(timePoint == -60) %>% # keep only fasting values
#   group_by(ID) %>% # calculate index for each subject
#   summarise(InsuSen = ISI_fasting(glu = glucose_mgdl, ins = insulin_microU,
#                                   nefa = nefa_umolL, index = "Revised QUICKI"))

Since our sample dataset also includes values from a 2-hour, 5-sample oral glucose tolerance test (OGTT), we might also go ahead and calculate ISI based on that. I'll demonstrate this using for the NEFA-ISI index:

# OGTT_long %>%
#   filter(timePoint %in% c(0, 60, 120)) %>% # select only the relevant timepoints
#   arrange(timePoint) %>% # make sure the timepoints are correctly ordered
#   group_by(ID) %>% # calculate index for each subject
#   summarise(InsuSen = ISI_OGTT(ins = insulin_pmoll, nefa = nefa_umolL,
#                                bmi = bmi[1], index = "NEFA-ISI")) # bmi[1] ensures only one value
                                                                  # is passed on to the function,
                                                                  # instead of a vector of
                                                                  # repeating values.

And so, as I previously demonstrated for fasting values, we obtain a data.frame with a column InsuSen (you can use whatever name you wish) with a NEFA-ISI value for each subject.

Below another example involving a little more data wrangling to calculate the Stumvoll-ISI index:

# OGTT_long %>%
#   gather(parameter, value, c(glucose_mgdl:insulin_microU)) %>% # gather all the columns into a single one
#   filter((timePoint == 90 & parameter == "glucose_mgdl") | (timePoint == 120 & parameter == "insulin_microU")) %>% # select only the relevant timepoints
#   group_by(ID) %>% # calculate index for each subject
#   summarise(InsuSen = ISI_OGTT(glu = value[parameter == "glucose_mgdl"],
#                                ins = value[parameter == "insulin_microU"],
#                                bmi = bmi[1], index = "Stumvoll-ISI"))