Errors of downsample and filter functions

[catsquito-dl]

Dear Dr. Forbes,

I obtained pupil data from HTC Vive Pro Eye and removed all 0 and -1 from both eyes. I followed the PupillometryR procedures here. I was able to run smoothing function, but when I run downsample I got the below error: I had the same error when running the median filter ( I guess it is because I didn't successfully downsample using option 'median'.)

This is how the data look like:

There are two subjects and four trials in the current data, would it be a problem? I am new to eye data analysis. Any suggestions will be appreciated!

[samhforbes]

Hi @catsquito-dl Do you have a small reproducible example you can post? It looks like perhaps the issue might be with the time bin size - but I can't tell until there's an example. What was your sample rate?

[catsquito-dl]

Hi @catsquito-dl Do you have a small reproducible example you can post? It looks like perhaps the issue might be with the time bin size - but I can't tell until there's an example. What was your sample rate?

Thanks for your reply. Sure I'd like to share an example, is it ok if I reach by email?

[samhforbes]

Yes of course, please feel free to send to the email on my GitHub profile!

[samhforbes]

Hi @catsquito-dl , I can't replicate this error. If I try the below it seems to run, although you should check a couple of things. Missing data seems to be -1000 in your data set, and also trial start time seems to be not the same across trials.

library(tidyverse)
library(PupillometryR)

data <- read_csv('F12_F06_T1T2_L1_pupil_deleteNA.csv')

hist(data$ET_PupilLeft_mm)
hist(data$ET_PupilRight_mm)

pupil_data <- data %>% 
  mutate(ET_PupilLeft_mm = ifelse(ET_PupilLeft_mm < 0, NA, ET_PupilLeft_mm),
         ET_PupilRight_mm = ifelse(ET_PupilRight_mm < 0, NA, ET_PupilRight_mm))

Sdata <- make_pupillometryr_data(data = pupil_data,
                                 subject = ID,
                                 trial = Trial,
                                 time = Timestamp,
                                 condition = Types)

plot(Sdata, pupil = ET_PupilLeft_mm, group = 'condition')

regressed_data <- regress_data(data = Sdata,
                               pupil1 = ET_PupilLeft_mm,
                               pupil2 = ET_PupilRight_mm)

mean_data <- calculate_mean_pupil_size(data = regressed_data,
                                       pupil1 = ET_PupilLeft_mm,
                                       pupil2 = ET_PupilRight_mm)

mean_data <- downsample_time_data(data = mean_data,
                                  pupil = mean_pupil,
                                  timebin_size = 50,
                                  option = 'median')

plot(mean_data, pupil = mean_pupil, group = 'condition')

[catsquito-dl]

Hi @catsquito-dl , I can't replicate this error. If I try the below it seems to run, although you should check a couple of things. Missing data seems to be -1000 in your data set, and also trial start time seems to be not the same across trials.

library(tidyverse)
library(PupillometryR)

data <- read_csv('F12_F06_T1T2_L1_pupil_deleteNA.csv')

hist(data$ET_PupilLeft_mm)
hist(data$ET_PupilRight_mm)

pupil_data <- data %>% 
  mutate(ET_PupilLeft_mm = ifelse(ET_PupilLeft_mm < 0, NA, ET_PupilLeft_mm),
         ET_PupilRight_mm = ifelse(ET_PupilRight_mm < 0, NA, ET_PupilRight_mm))

Sdata <- make_pupillometryr_data(data = pupil_data,
                                 subject = ID,
                                 trial = Trial,
                                 time = Timestamp,
                                 condition = Types)

plot(Sdata, pupil = ET_PupilLeft_mm, group = 'condition')

regressed_data <- regress_data(data = Sdata,
                               pupil1 = ET_PupilLeft_mm,
                               pupil2 = ET_PupilRight_mm)

mean_data <- calculate_mean_pupil_size(data = regressed_data,
                                       pupil1 = ET_PupilLeft_mm,
                                       pupil2 = ET_PupilRight_mm)

mean_data <- downsample_time_data(data = mean_data,
                                  pupil = mean_pupil,
                                  timebin_size = 50,
                                  option = 'median')

plot(mean_data, pupil = mean_pupil, group = 'condition')

Dear Dr. Forbes,

Thanks for the update! Sorry, yesterday I was out of town and couldn't reply.

I tried the code you posted and it works. I guess mutate() helps although I am unfamiliar with this function or dplyr package which I will learn more about it.

I also tried the window analyses, however, some mean_pupil from window is NaN. I assume this is caused by the inconsistent trial start time. There are missing data which are -1000, that's correct. Previously, I manually deleted these data because I am afraid these -1000 would cause errors.

I will fix the trial start time first and see if I can make window analyses work.

[catsquito-dl]

Dear Dr. Forbes,

I am trying to display the values of mean and quartiles in the raincloud plot in the window analysis, but it sees not to work. Could you help with it?

base_data <- baseline_data(data = filtered_data, pupil = mean_pupil, start = 0, stop = 35000)

window <- create_window_data(data = base_data, pupil = mean_pupil) plot(window, pupil = mean_pupil, windows = FALSE, geom = 'raincloud', for (i in unique(window)) { mean_val <- round(mean(mean_pupil$pupil[window == i]), 2) text(x = i, y = 0.5, labels = mean_val, col = "red") })

[catsquito-dl]

Below is the results of window analysis:

and this is the plot:

I manually calculated the mean values of two conditions (baseline, challenging), but in the results mean of baseline > mean of challenging, which is different from the raincloud plot. I am confused. Any suggestions will be appreciated.

[samhforbes]

Hi @catsquito-dl I only have the data you sent previously which is just 2 participants, so I can't check this. If you'd like to send updated data I can take a look.

Are those baseline values the ones you want to use? And what do you want the plot to look like? I can't visualise that in my head. Note the boxplot already gives IQR by default.

[samhforbes]

Closing as inactive, but feel free to reopen if something comes up.