A simple way to group timestamped data by an arbitrary maximum amount of time between events. Inspired by this (http://stackoverflow.com/a/32248462) Stack Overflow answer.

identifying_sessions.R

# Load helpful packages
library(lubridate)
library(dplyr)

# Sample Data
data <- structure(list(X1 = c("kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov","kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "atl1.america.net", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "atl1.america.net", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "atl1.america.net", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov", "kenneth.lerc.nasa.gov"), 
                       X2 = structure(c(807272161, 807892754, 807892821, 807892978, 807893004, 807956939, 807957024, 807957075, 807958536, 807958583, 807958612, 807958613, 808017688, 808151728, 808151755, 808151885, 808151897, 808151902, 808151914, 808151930, 808151934, 808151942, 808151950, 808151966, 808151987, 808152076, 808152077, 808152211, 808152240, 808152278, 808443246, 808647297, 808647305, 808648062, 808648062, 808648067, 808648291, 809002676, 809002681, 809002692, 809002794, 809518045, 809518054, 809518076, 809518084, 809518165, 809518178, 809518190, 809518198, 809518935, 809519059, 809519138, 809519152, 809519192, 809519202, 809519310, 809519321, 809519331, 809519383, 809519400, 809519623, 809519677, 809519774, 809519784, 809519837, 809606867, 809606878, 809606884, 809618574, 809618579, 809618584, 809693098, 809770817, 809770822, 809770827, 809770868, 809771523), 
                                      tzone = "UTC", 
                                      class = c("POSIXct", "POSIXt"))), 
                  class = c("tbl_df", "tbl", "data.frame"), 
                  row.names = c(NA, -77L), 
                  .Names = c("X1", "X2"))

# What this code does is:
# 1. Group by some variable (in this case website, in your case patient ID)
# 2. Create "diff_time" which is the current timestamp minus the previous (lag) timestamp - there is a bug right now that isn't letting me set the lag() default as minimum timestamp which means we have to deal with the first diff_time being NA.
# 3. Create a flag variable that puts the number 1 everywhere that the diff_time is larger than the time interval you desire - here it is 2 minutes. If the original time interval was NA set it equal to 1
# 4. Assign Session ID based on the cumulative sum of the flag variables
data %>% 
  group_by(X1) %>% 
  mutate(diff_time = X2 - lag(X2, order_by = X2),
         flag = ifelse(is.na(diff_time), 
                       yes = 1, 
                       no = ifelse(diff_time > seconds(120), 
                                   yes = 1, 
                                   no = 0)),
         session_id =  cumsum(flag))

Ok in a different version of dplyr this code works:

data %>% 
  group_by(X1) %>% 
  arrange(X1, X2) %>%
  mutate(diff_time = X2 - lag(X2, order_by = X2, default = min(X2)),
        flag = ifelse(diff_time > seconds(120), yes = 1, no = 0), 
        session_id =  cumsum(flag)+1)

NOTE: You must have the arrange step for the cumsum() to work appropriately!!!!!!!!!!!!!!!!