I was wondering if someone could help me with specificity/sensitivity code. I have a database with 500 patients and I have a column called age_group with 4 different age groups. I have a test and a disease and I would like to look at sens/spec for the test to predict this specific disease.
I used tidyverse, finalfit and yardstick. Disease 1, 0. Test ranges values are from about 5 to 12. 4 age groups.
Hi all, I was wondering if someone could help me with specificity/sensitivity code. I have a database with 500 patients and I have a column called age_group with 4 different age groups. I have a test and a disease and I would like to look at sens/spec for the test to predict this specific disease.
I used tidyverse, finalfit and yardstick. Disease 1, 0 Test ranges values are from about 5 to 12 4 age groups
This is the code I used:
prognos <- function(.data, estimate, truth){
.estimate = sym (estimate)
.truth = sym (truth)
.data %>%
drop_na(!! .estimate, !!.truth) %>%
summarise(
N = n(),
TP = sum (!! .truth == 1 & !!.estimate ==1),
TN = sum (!!.truth == 0 & !!.estimate == 0),
FP = sum (!!.truth == 0 & !!.estimate == 1),
FN = sum (!!.truth ==1 & !!.estimate == 0),
Sens = sens (.,!!.truth, !!.estimate)$.estimate,
Spec = spec (.,!!.truth, !!.estimate)$.estimate,
PPV = ppv (.,!!.truth, !!.estimate)$.estimate,
NPV = npv (.,!!.truth, !!.estimate)$.estimate,
above = paste0(TP +FP, " (", ((100*(TP + FP))/N) %>% round_tidy (1), ")"),
below = paste0(TN+ FN, "(", ((100*(TN + FN))/N) %>% round_tidy (1), ")")
) %>%
mutate_at(vars("Sens", "Spec", "PPV", "NPV"), ~ prod(.,100)) %>%
mutate(across(c(everything(), -above, -below), round, digits=1))
}
database %>%
prognos(estimate = "test", truth = "disease")
Unfortunately this code doesn't work if I add groupby(agegroup). It gives out a tibble of 16x4 with 4 lines for each age groups. Sens/spec the same in the whole tibble.
Like this:
Age_group N TP TN FP Sens Spec PPV NPV
40-49 61 11 31 4 12.6 30.8 15.2 15
40-49 61 11 31 4 12.6 30.8 15.2 15
40-49 61 11 31 4 12.6 30.8 15.2 15
40-49 61 11 31 4 12.6 30.8 15.2 15
50-59 111 33 36 27 12.6 30.8 15.2 15
50-59 111 33 36 27 12.6 30.8 15.2 15
etc
Does anyone know how this can be fixed and how can I get accurate sens/spec for each age group?
Thank you.