0

I created a multilevel regression model with nlme package and now I would like to illustrate the regression line obtained for some patients (unfortunately I cannot use geom_smooth with nlme).

So using the model I obtained the following predicted values (predicted_value) at different times (date_day) and here for two patients (ID1 and ID2).

df <- data.frame (ID = c (rep (1, 10), rep(2, 10)),
                  date_day = c (7:16, 7:16),
                  predicted_value = c (33, 33, 33, 33, 33, NA, 34, NA, NA, NA, 
                               55, NA, NA, 53.3, NA, NA, 51.6, NA, 50.5, NA))


 ID date_day predicted_value
1   1        7            33.0
2   1        8            33.0
3   1        9            33.0
4   1       10            33.0
5   1       11            33.0
6   1       12              NA
7   1       13            34.0
8   1       14              NA
9   1       15              NA
10  1       16              NA
11  2        7            55.0
12  2        8              NA
13  2        9              NA
14  2       10            53.3
15  2       11              NA
16  2       12              NA
17  2       13            51.6
18  2       14              NA
19  2       15            50.5
20  2       16              NA

Now I would like to draw the regression line for each of these patients. So I tried the following

ggplot(df%>% filter(ID %in% c("1", "2")))+ 
  aes(x =  date_day, y = predicted_value) +
  geom_point(shape = "circle", size = 1.5, colour = "#112446", na.rm = T) +
  geom_line(aes(y = predicted_value), na.rm = T, size = 1) +
  theme_minimal() +
  facet_wrap(vars(ID)) +
  scale_x_continuous(name="days", limits=c(7, 16)) +
  scale_y_continuous(name="predicted values", limits=c(0, 60))

But I end with the following plots: patient 1 : the line is interrupted, and patient 2 no line at all. How can I fix that ?

enter image description here

Thanks a lot

David
  • 111
  • 6
  • 1
    your problem is the `NA` values in the predicted values. In this particular case I think just using `na.omit(df)` should do what you want ... ? – Ben Bolker Mar 22 '22 at 16:12
  • Thank you, indeed changing ggplot(df%>% filter(ID %in% c("1", "2"))) for ggplot(na.omit(df)%>% filter(ID %in% c("1", "2"))) allowed to solve the pb – David Mar 22 '22 at 16:19
  • 1
    You're encouraged to post the solution to your own problem (you may have to wait an hour or so, I don't remember) – Ben Bolker Mar 22 '22 at 16:20

1 Answers1

1

Thank you @BenBolker , indeed changing the first line

ggplot(df%>% filter(ID %in% c("1", "2")))

to

ggplot(na.omit(df)%>% filter(ID %in% c("1", "2")))

allowed to solve the job

Ben Bolker
  • 211,554
  • 25
  • 370
  • 453
David
  • 111
  • 6