Hello list,
I want to make a large rulebased algorithm, to provide decision support for
drug prescriptions. I have defined the algorithm in a function, with a for loop
and many if statements. The structure should be as follows:
1. Iterate over a list of drug names. For each drug:
2. Get some drug related data (external dataset). Row of a dataframe.
3. Check if adaptions should be made to standard dosage and safety information
in case of contraindications. If patient has an indication, update current
dosage and safety information with the value from the dataframe row.
4. Save dosage and safety information in some lists and continue to the next
drug.
5. When the iteration over all drugs is done, return the lists.
ISSUE:
So it is a very large function with many nested if statements. I have checked
the code structure multiple times, but i run into some issues. When i try to
run the function definiton, the command never "completes" in de console.
Instead of ">", the console shows "+". No errors are raised.
As I said, i have checked the structure multiple times, but cant find an error.
I have tried rebuilding it and testing each time i add a part. Each part
functions isolated, but not together in the same function. I can't find any
infinite loops either.
I suspect the function may be too large, and i have to define functions for
each part separately. That isn't an issue necessarily, but i would still like
to know why my code won't run. And whether there are any downsides or
considerations for using many small functions.
Below is my code. I have left part of it out. There are six more parts like the
diabetes part that are similar.
I also use a lot of data/variabeles not included here, to try and keep things
compact. But I can provide additional information if helpful.
Thanks it advance for thinking along!!
Kind regards,
Emily
The code:
decision_algorithm <- function(AB_list, dataset_ab = data.frame(), diagnose =
'cystitis', diabetes_status = "nee", katheter_status = "nee",
lang_QT_status = "nee", obesitas_status = "nee",
zwangerschap_status = "nee",
medicatie_actief = data.frame(dict[["med_AB"]]),
geslacht = "man", gfr=90){
# vars
list_AB_status <- setNames(as.list(rep("green", length(AB_list))),
names(AB_list)) #make a dict of all AB's and assign status green as deafault
for status
list_AB_remarks <- setNames(as.list(rep("Geen opmerkingen",
length(AB_list))), names(AB_list)) #make a dict of all AB's and assign "Geen"
as default for remarks #Try empty list
list_AB_dosering <- setNames(as.list(rep("Geen informatie",
length(AB_list))), names(AB_list)) # make named list of all AB's and assign
"Geen informatie", will be replaced with actual information in algorithm
list_AB_duur <- setNames(as.list(rep("Geen informatie", length(AB_list))),
names(AB_list)) # make named list of all AB's and assign "Geen informatie",
will be replaced with actual information in algorithm
##### CULTURES #####
for (i in names(AB_list)) {
ab_data <- dataset_ab[dataset_ab$middel == i,] #get info for this AB from
dataset_ab
# Extract and split the diagnoses, dosering, and duur info for the current
antibiotic
ab_diagnoses <- str_split(ab_data$diagnoses, pattern = " \\| ")[[1]]
ab_diagnose_dosering <- str_split(ab_data$`diagnose dosering`, pattern = "
\\| ")[[1]]
ab_diagnose_duur <- str_split(ab_data$`diagnose duur`, pattern = " \\|
")[[1]]
# Find the index of the current diagnose in the ab_diagnoses list
diagnose_index <- match(diagnose, ab_diagnoses)
# Determine dosering and duur based on the diagnose_index
if (!is.na(diagnose_index)) {
dosering <- ifelse(ab_diagnose_dosering[diagnose_index] == "standaard",
ab_data$dosering, ab_diagnose_dosering[diagnose_index])
duur <- ifelse(ab_diagnose_duur[diagnose_index] == "standaard",
ab_data$duur, ab_diagnose_duur[diagnose_index])
} else {
# Use general dosering and duur as fallback if diagnose is not found
dosering <- ab_data$dosering
duur <- ab_data$duur
}
list_AB_dosering[[i]] <- dosering
list_AB_duur[[i]] <- duur
if ((!is.null(AB_list[[i]]) && AB_list[[i]] == "I")) {
list_AB_status[[i]] <- "yellow"
list_AB_remarks[[i]] <- "Kweek verminderd gevoelig"
} else if ((!is.null(AB_list[[i]]) && AB_list[[i]] == "R")) {
list_AB_status[[i]] <- "red"
list_AB_remarks[[i]] <- "Kweek resistent"
}else if ((!is.null(AB_list[[i]]) && AB_list[[i]] == "S")) {
next
} else {
list_AB_status[[i]] <- "yellow"
list_AB_remarks[[i]] <- "Geen kweekgegevens"
}
# counters, for check if dosering / duur are updated more than once
dosering_update_count <- 0
duur_update_count <- 0
##### DIABETES #####
if (diabetes_status == "ja") {
if (ab_data$'diabetes veiligheid' == "ja") {
list_AB_status[[i]] <- "red"
list_AB_remarks[[i]] <- paste(list_AB_remarks[[i]], "Niet veilig met
diabetes")
}
if (ab_data$'diabetes effectiviteit' == "aanpassing") {
dosering <- ifelse(ab_data$'diabetes dosering' != "standaard",
ab_data$'diabetes dosering', dosering) # if dosering does not equal standaard,
apply dosering in column, otherwise keep initial dosering
duur <- ifelse(ab_data$'diabetes duur' != "standaard",
ab_data$'diabetes duur', duur) # if dosering does not equal standaard, apply
dosering in column, otherwise keep initial dosering
dosering_update_count <- dosering_update_count + 1
duur_update_count <- duur_update_count + 1
list_AB_remarks[[i]] <- paste(list_AB_remarks[[i]], ab_data$'diabetes
opmerkingen')
}
} else if (diabetes_status == "?") {
if (ab_data$'diabetes veiligheid' == "ja") {
list_AB_remarks[[i]] <- paste(list_AB_remarks[[i]], "Waarschuwing: Dit
middel kan veiligheidsimplicaties hebben bij diabetes.")
}
if (ab_data$'diabetes effectiviteit' == "aanpassing") {
list_AB_remarks[[i]] <- paste(list_AB_remarks[[i]], "Waarschuwing: Dit
middel kan dosisaanpassingen vereisen bij diabetes.")
}
}
list_AB_dosering[[i]] <- dosering
list_AB_duur[[i]] <- duur
# within for loop
}
# within function
return(list(status = list_AB_status, remarks = list_AB_remarks, duur =
list_AB_duur, dosering = list_AB_dosering))
}
______________________________________________
[email protected] mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
