如何计算每天的出院率和人数？

Question

我想知道如何计算每天的患者人数，它想包括当天的患者出院，但有入院患者。 因此，7/17 不应找到患者 c。

我有一个更大的数据集。这只是一个例子。

谢谢你帮助我。

install.packages("lubridate")
library(lubridate)
admission <- c("06/23/2013", "06/30/2013", "07/12/2014","06/24/2013","06/28/2013","06/29/2013","06/23/2013","06/24/2013","06/24/2013")
discharge<- c("06/25/2013", "07/03/2013", "07/17/2014","06/30/2013","06/30/2013","07/02/2013","06/29/2013","06/29/2013","06/27/2013")
patient<-c("a","b","c","d","e","f","g","h","j")
admission.date <- mdy(admission)
discharge.date <- mdy(discharge)
df<-data.frame(patient,admission.date,discharge.date)
df
  patient admission.date discharge.date
1       a     2013-06-23    2013-06-25
2       b     2013-06-30    2013-07-02
3       c     2014-07-12    2014-07-17
4       d     2013-06-24    2013-06-30
5       e     2013-06-28    2013-06-30
6       f     2013-06-29    2013-07-02
7       g     2013-06-23    2013-06-29
8       h     2013-06-24    2013-06-29
9       j     2013-06-24    2013-06-27
  

Answer 1

这是一种使用data.table的方法

library(data.table)
# set df to data.table format
setDT(df)
# Create a table with all dates
dt.dates <- data.table( date = seq(min(df$admission.date), max(df$discarge.date), by = "1 days") )
#  perform overlap join
answer <- df[dt.dates, .(date, patient), on = .(admission.date <= date, discarge.date > date), nomatch = 0L]
# get unique patients by date
answer[, .(patients = uniqueN(patient)), by = date]
#          date patients
# 1: 2013-06-23        2
# 2: 2013-06-24        5
# 3: 2013-06-25        4
# 4: 2013-06-26        4
# 5: 2013-06-27        3
# 6: 2013-06-28        4
# 7: 2013-06-29        3
# 8: 2013-06-30        2
# 9: 2013-07-01        2
#10: 2013-07-02        1
#11: 2014-07-12        1
#12: 2014-07-13        1
#13: 2014-07-14        1
#14: 2014-07-15        1
#15: 2014-07-16        1

Answer 2

这是一种使用dplyr的方法：

library(dplyr)

df %>%
  rowwise() %>%
  do(data.frame(patient=.$patient, Date=seq(.$admission.date,.$discarge.date-1,by="day"))) %>%
  group_by(Date) %>%
  summarize(patients = n())

有输出：

# A tibble: 15 x 2
   Date       patients
   <date>        <int>
 1 2013-06-23        2
 2 2013-06-24        5
 3 2013-06-25        4
 4 2013-06-26        4
 5 2013-06-27        3
 6 2013-06-28        4
 7 2013-06-29        3
 8 2013-06-30        2
 9 2013-07-01        2
10 2013-07-02        1
11 2014-07-12        1
12 2014-07-13        1
13 2014-07-14        1
14 2014-07-15        1
15 2014-07-16        1