processing of hospital admission data using R - r

Question

I have a set of hospital admission data that I need to process, I am stuck when trying to loop the data and pick up the stuff I need, here is the example:
Date Ward
1 A
2 A
3 A
4 A B
5 A
6 A
7 A C
8 C
9 C
10 C
And I need them to be transformed into:
Ward Adm_Date Dis_Date
A 1 4
B 4 4
A 4 7
C 7 10
To put it in sentence, this is a admission record patient X who:
go to ward A from day 1 to day 4
go to ward B (maybe it's an ICU ward) for less than a day in day 4, and move back to ward A on that day
stay in ward A from day 4 to day 7
move to ward C from ward A from day 7 and stay in ward C till day 10
I am thinking of using ddply by filtering the ward but it is not OK since B will be "omitted" and the period of time for A is not broken down into 2 pieces.
Any suggestions? Thanks!

Answer 1

dat <- data.frame(Date=1:10,Ward=c(rep("A",3),"A B",rep("A",2),"A C",rep("C",3)))
dat$Ward <- as.character(dat$Ward)
# Change data to a "long" format
Date2 <- rep(dat$Date,nchar(gsub(" ","",dat$Ward)))
Ward2 <- unlist(strsplit(dat$Ward," "))
dat2 <- data.frame(Date=Date2,Ward=Ward2)
dat2$Ward <- as.character(dat2$Ward) # pesky factors!
# Create output
Ward3 <- unlist(strsplit(gsub("(\\w)\\1+","\\1",paste(dat2$Ward,collapse="")),""))
#helper function to find lengths of repeated characters, probably a better way of doing this
repCharLength <- function(str)
{
out <- numeric(0)
tmp <- 1
for (i in 2:length(str))
{
if (str[i]!=str[i-1])
{out<-c(out,tmp)
tmp<-1}
else
tmp <- tmp+1
}
return(c(out,tmp))
}
stays <- repCharLength(dat2$Ward)
Adm_Date <- c(1,dat2$Date[cumsum(stays)[1:(length(stays)-1)]])
Dis_Date <- dat2$Date[cumsum(stays)]
dat3 <- data.frame(Ward=Ward3,Adm_Date=Adm_Date,Dis_Date=Dis_Date)
> dat3
Ward Adm_Date Dis_Date
1 A 1 4
2 B 4 4
3 A 4 7
4 C 7 10
A bit more involved than I first thought, and there is probably a better way to get the stay lengths than using the helper function I wrote, but this seems to do the job.
Edit
In light of Spacedman's comment, there is a library function to calculate Ward3 and stays:
Ward3 <- rle(dat2$Ward)$values
stays <- rle(dat2$Ward)$lengths

Answer 2

It's not a complex answer but you can transform your data
X <- data.frame(
Date=1:10,
Ward=c("A","A","A","A B","A","A","A C","C","C","C"),
stringsAsFactors=FALSE
)
w <- strsplit(X$Ward," +")
n <- sapply(w, length)
X_mod <- data.frame(
Date = rep(X$Date, n),
Ward = unlist(w, FALSE, FALSE)
)
With X_mod you could write vectorized (=fast) solution. For start with(X_mod, c(0,cumsum(Ward[-1]!=Ward[-length(Ward)]))) gives you id of visit.