readxl::read_excel(here("PIG PAUS data_yc.xlsx")) %>%
janitor::clean_names() |>
janitor::remove_empty() |>
mutate(measurement_value = as.numeric(measurement_value)) ->
dfvalue for "which" not specified, defaulting to c("rows", "cols")Warning: There was 1 warning in `mutate()`.
ℹ In argument: `measurement_value =
as.numeric(measurement_value)`.
Caused by warning:
! NAs introduced by coerciondf |>
group_by(pig, treated, wavelength) |>
summarise(mean = mean(measurement_value, na.rm=TRUE),
sd = sd(measurement_value, na.rm=TRUE)) |>
gt() |>
tab_header(
title = md("Data Grouped by Pig, Treatment, and Wavelength")
)`summarise()` has grouped output by 'pig', 'treated'. You
can override using the `.groups` argument.Data Grouped by Pig, Treatment, and Wavelength |
||
|---|---|---|
| wavelength | mean | sd |
| PAPC-1 - 0 | ||
| 800 | 4283.677 | 423.8885 |
| 1200 | 3634.495 | 301.0695 |
| PAPC-1 - 1 | ||
| 800 | 6255.793 | 203.0605 |
| 1200 | 5340.887 | 573.8289 |
| PAPC-3 - 0 | ||
| 750 | 4633.550 | 143.4739 |
| 800 | 4625.280 | 309.0905 |
| 1200 | 3810.645 | 528.7532 |
| 1310 | 3709.348 | 535.2551 |
| PAPC-3 - 1 | ||
| 800 | 6751.210 | 450.3719 |
| 1200 | 6448.237 | 622.0233 |
| 1310 | 5500.300 | 121.5375 |
df |>
group_by(treated, wavelength) |>
summarise(mean = mean(measurement_value, na.rm=TRUE),
sd = sd(measurement_value, na.rm=TRUE)) |>
gt() |>
tab_header(
title = md("Data Grouped by Treatment (0/1) and Wavelength")
)`summarise()` has grouped output by 'treated'. You can
override using the `.groups` argument.Data Grouped by Treatment (0/1) and Wavelength |
||
|---|---|---|
| wavelength | mean | sd |
| 0 | ||
| 750 | 4633.550 | 143.4739 |
| 800 | 4369.078 | 408.6534 |
| 1200 | 3678.533 | 333.6674 |
| 1310 | 3709.348 | 535.2551 |
| 1 | ||
| 800 | 6503.502 | 413.8341 |
| 1200 | 5894.562 | 808.9155 |
| 1310 | 5500.300 | 121.5375 |
We can perform unpaired t tests without regard to pig or esophageal location
df |>
filter(wavelength == 800) |>
rstatix::t_test(measurement_value ~ treated, var.equal = TRUE) # A tibble: 1 × 8
.y. group1 group2 n1 n2 statistic df p
* <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl>
1 measure… 0 1 8 6 -9.62 12 5.43e-7df |>
filter(wavelength == 1200) |>
rstatix::t_test(measurement_value ~ treated, var.equal = TRUE) # A tibble: 1 × 8
.y. group1 group2 n1 n2 statistic df p
* <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl>
1 measure… 0 1 8 6 -7.06 12 1.32e-5df |>
filter(wavelength ==1310) |>
rstatix::t_test(measurement_value ~ treated, var.equal = TRUE) # A tibble: 1 × 8
.y. group1 group2 n1 n2 statistic df p
* <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl>
1 measure… 0 1 5 2 -4.44 5 0.00675There appear to be significant differences between treated and untreated pigs at all wavelengths. The differences are most pronounced at 800 nm.
model <- lm(measurement_value ~ pig + treated + site_cm, data = df |> filter(wavelength == 800))
model |>
broom::tidy() # A tibble: 4 × 5
term estimate std.error statistic p.value
<chr> <dbl> <dbl> <dbl> <dbl>
1 (Intercept) 4047. 498. 8.12 0.0000103
2 pigPAPC-3 411. 222. 1.85 0.0939
3 treated 2020. 215. 9.38 0.00000284
4 site_cm 5.77 12.6 0.456 0.658 Even with this small N, there is some variance from pig to pig at 800 nm. There is a trend toward higher values for Pig 3. There is a significant difference between treated and untreated pigs. Treatment increases signal at 800 nm by ~ 2020 AU. There is minimal difference between the different esophageal sites (with this small N). We are not controlling for pig age, which might correlate with higher signals.
model <- lm(measurement_value ~ pig + treated + site_cm, data = df |> filter(wavelength == 1200))
model |>
broom::tidy() # A tibble: 4 × 5
term estimate std.error statistic p.value
<chr> <dbl> <dbl> <dbl> <dbl>
1 (Intercept) 4014. 670. 5.99 0.000134
2 pigPAPC-3 659. 298. 2.21 0.0516
3 treated 2078. 289. 7.18 0.0000301
4 site_cm -13.2 17.0 -0.775 0.457 Even with this small N, there is some variance from pig to pig at 1200 nm. There is a trend toward higher values for Pig 3. There is a significant difference between treated and untreated pigs. Treatment increases signal at 1200 nm by ~ 2078 AU. There is minimal difference between the different esophageal sites (with this small N). We are not controlling for pig age, which might correlate with higher signals.
model <- lm(measurement_value ~ treated + site_cm,
data = df |> filter(wavelength == 1310))
model |>
broom::tidy() # A tibble: 3 × 5
term estimate std.error statistic p.value
<chr> <dbl> <dbl> <dbl> <dbl>
1 (Intercept) 3181. 1238. 2.57 0.0620
2 treated 1857. 466. 3.99 0.0163
3 site_cm 13.2 30.4 0.434 0.686 Only one pig was evaluated at 1310 nm. There is a significant difference between treated and untreated pigs.Treatment increases signal at 1310 nm by ~ 1857 AU. There is minimal difference between the different esophageal sites (with this small N). We are not controlling for pig age, which might correlate with higher signals.
First, let’s look at the differencs at 800 nm
df |>
filter(wavelength == 800) |>
arrange(pig, site_cm, treated) |>
filter(lag(site_cm) == site_cm | lead(site_cm) == site_cm) |>
mutate(diff = measurement_value - lag(measurement_value)) |>
filter(treated ==1) |>
gt()| pig | date | site_cm | wavelength | treated | measurement_value | diff |
|---|---|---|---|---|---|---|
| PAPC-1 | 2024-06-25 | 30 | 800 | 1 | 6162.62 | 1612.88 |
| PAPC-1 | 2024-06-25 | 40 | 800 | 1 | 6488.72 | 1703.53 |
| PAPC-1 | 2024-06-25 | 50 | 800 | 1 | 6116.04 | 1463.56 |
Now let’s perform the paired t-test at 800 nm - in two ways, first by looking at the differences in the same pig at the same site, then by using the official paired t test.
df |>
filter(wavelength == 800) |>
arrange(pig, site_cm, treated) |>
filter(lag(site_cm) == site_cm | lead(site_cm) == site_cm) |>
mutate(diff = measurement_value - lag(measurement_value)) |>
filter(treated ==1) |>
rstatix::t_test(diff ~ 1, mu = 0) # A tibble: 1 × 7
.y. group1 group2 n statistic df p
* <chr> <chr> <chr> <int> <dbl> <dbl> <dbl>
1 diff 1 null model 3 22.8 2 0.00192df |>
filter(wavelength == 800) |>
arrange(pig, site_cm, treated) |>
filter(site_cm %in% c(30,40,50)) |>
filter(pig == 'PAPC-1') |>
rstatix::t_test(measurement_value ~ treated, paired = TRUE)# A tibble: 1 × 8
.y. group1 group2 n1 n2 statistic df p
* <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl>
1 measure… 0 1 3 3 -22.8 2 0.00192Again, the differences with treatment are significant.
First, let’s look at the differencs at 800 nm
df |>
filter(wavelength == 800) |>
arrange(pig, site_cm, treated) |>
filter(lag(site_cm) == site_cm | lead(site_cm) == site_cm) |>
mutate(diff = measurement_value - lag(measurement_value)) |>
filter(treated ==1) |>
gt()| pig | date | site_cm | wavelength | treated | measurement_value | diff |
|---|---|---|---|---|---|---|
| PAPC-1 | 2024-06-25 | 30 | 800 | 1 | 6162.62 | 1612.88 |
| PAPC-1 | 2024-06-25 | 40 | 800 | 1 | 6488.72 | 1703.53 |
| PAPC-1 | 2024-06-25 | 50 | 800 | 1 | 6116.04 | 1463.56 |
Now let’s perform the paired t-test at 1200 nm - in two ways, first by looking at the differences in the same pig at the same site, then by using the official paired t test.
df |>
filter(wavelength == 1200) |>
arrange(pig, site_cm, treated) |>
filter(lag(site_cm) == site_cm | lead(site_cm) == site_cm) |>
mutate(diff = measurement_value - lag(measurement_value)) |>
filter(treated ==1) |>
rstatix::t_test(diff ~ 1, mu = 0) # A tibble: 1 × 7
.y. group1 group2 n statistic df p
* <chr> <chr> <chr> <int> <dbl> <dbl> <dbl>
1 diff 1 null model 3 3.57 2 0.0702df |>
filter(wavelength == 1200) |>
arrange(pig, site_cm, treated) |>
filter(site_cm %in% c(30,40,50)) |>
filter(pig == 'PAPC-1') |>
rstatix::t_test(measurement_value ~ treated, paired = TRUE) # A tibble: 1 × 8
.y. group1 group2 n1 n2 statistic df p
* <chr> <chr> <chr> <int> <int> <dbl> <dbl> <dbl>
1 measurem… 0 1 3 3 -3.57 2 0.0702Again, the differences with treatment are significant.
Now let’s perform the paired t-test at 1310 nm.
df |>
filter(wavelength == 1310) |>
arrange(pig, site_cm, treated) |>
filter(lag(site_cm) == site_cm | lead(site_cm) == site_cm) |>
mutate(diff = measurement_value - lag(measurement_value)) |>
filter(treated ==1) |>
rstatix::t_test(diff ~ 1, mu = 0) # A tibble: 1 × 7
.y. group1 group2 n statistic df p
* <chr> <chr> <chr> <int> <dbl> <dbl> <dbl>
1 diff 1 null model 2 16.2 1 0.0392Again, the differences with treatment are significant.
df |>
ggplot(aes(x = site_cm, y = measurement_value, color = factor(treated))) +
geom_point() +
facet_grid(pig ~ wavelength) +
labs(title = "Measurement values by Wavelength, Pig, and APC treatment",
x = "Site (cm)",
y = "Measurement value (AU)",
color = "Treated") +
scale_color_manual(values = c("black", "red"),
labels = c("No Rx", "APC"),
guide = guide_legend(reverse = TRUE)) +
theme_minimal(base_size = 16)Warning: Removed 1 row containing missing values or values outside
the scale range (`geom_point()`).