Update Survival Plots

[gowerc]

Closes #200

@danielinteractive - Still a work in progress but just wanted to share my current state to get any initial feedback. Currently I've delegated the process of generating the data to a dedicated predict function so that users can return the underlying data if they want it e.g:

survival_samples <- survival(stan_samples)

pts <- sample(dat_os$pt, 4)
predict(
    survival_samples,
    patients = pts
)

    time       median        lower        upper    group type
1      0 1.000000e+00 1.000000e+00 1.000000e+00 pt_00099 surv
2     10 9.734132e-01 9.692208e-01 9.769738e-01 pt_00099 surv
3     20 9.475333e-01 9.393889e-01 9.544777e-01 pt_00099 surv
4     30 9.223415e-01 9.104752e-01 9.324998e-01 pt_00099 surv
5     40 8.978197e-01 8.824514e-01 9.110277e-01 pt_00099 surv
6     50 8.739497e-01 8.552899e-01 8.900500e-01 pt_00099 surv
7     60 8.507142e-01 8.289654e-01 8.695555e-01 pt_00099 surv
8     70 8.280962e-01 8.034502e-01 8.495328e-01 pt_00099 surv
9     80 8.060794e-01 7.787204e-01 8.299712e-01 pt_00099 surv
...

## Also supports groupings
pts <- list(
    "g1" = sample(dat_os$pt, 3),
    "g2" = sample(dat_os$pt, 5),
    "g3" = sample(dat_os$pt, 10)
)

predict(
    survival_samples,
    patients = pts,
    type = "haz",
    time_grid = c(0, 100, 200)
)

  time      median       lower       upper group type
1    0 0.003011690 0.002616886 0.003464048    g1  haz
2  100 0.003011690 0.002616886 0.003464048    g1  haz
3  200 0.003011690 0.002616886 0.003464048    g1  haz
4    0 0.006655227 0.005623280 0.007806013    g2  haz
5  100 0.006655227 0.005623280 0.007806013    g2  haz
6  200 0.006655227 0.005623280 0.007806013    g2  haz
7    0 0.004403979 0.003879670 0.004916867    g3  haz
8  100 0.004403979 0.003879670 0.004916867    g3  haz
9  200 0.004403979 0.003879670 0.004916867    g3  haz

The autoplot function is now essentially just a thin wrapper around this e.g.

pts <- list(
    "g1" = sample(dat_os$pt, 4),
    "g2" = sample(dat_os$pt, 4)
)

autoplot(
    survival_samples,
    pts,
    type = "cumhaz",
    add_wrap = FALSE
)

I still need to re-implement the KM option and need to update all the unit tests accordingly.

[danielinteractive]

Thanks @gowerc yeah sounds good!

[gowerc]

@danielinteractive - Hmm only thing is it looks like the ggplot2.utils only supports survival estimates for the KM curves.

Options are:

1. Restrict this feature for type="surv" only
2. Code something up by hand so that we can support type="cumhaz" as well (not too difficult)

Though for "haz" and "loghaz" I'm not so sure how to proceed. Apparently you can get empirical estimates using muhaz::muhaz (link) though it would not be accurate to call these "km" so would perhaps need to rename the argument add_reference or something like that ?

[danielinteractive]

We could also add these features to ggplot2.utils (because it is our package). But I would say for now we can restrict to km and put the rest on backlog

[gowerc]

Makes sense. I've put a backlog item for us here and a enhancement ticket for ggplot2.utils here

[gowerc]

Ok I think this PR is now feature complete. I just need to do the documentation and unit tests...

pts <- list(
    "g1" = sample(dat_os$pt, 100),
    "g2" = sample(dat_os$pt, 100)
)

autoplot(
    survival_samples,
    pts,
    add_km = TRUE
)

[gowerc]

Hey @danielinteractive , This is now ready for review. Apologies for the delay the documentation and testing ended up being more involved that I was expecting. That being said it should be all ready now (pending any bugs found by the cicd testing)

[github-actions]

Code Coverage Summary

Filename                       Stmts    Miss  Cover    Missing
---------------------------  -------  ------  -------  ----------------------
R/DataJoint.R                     27       0  100.00%
R/DataLongitudinal.R              71       0  100.00%
R/DataSurvival.R                  46       0  100.00%
R/defaults.R                      11       6  45.45%   27, 57-84, 111
R/generics.R                      11       1  90.91%   173
R/JointModel.R                    46       5  89.13%   85-99
R/JointModelSamples.R             54       0  100.00%
R/Link.R                           6       0  100.00%
R/LinkGSF.R                       63      12  80.95%   109-120
R/LinkNone.R                       3       1  66.67%   36
R/LinkRandomSlope.R               10       0  100.00%
R/LongitudinalGSF.R               23       0  100.00%
R/LongitudinalModel.R             10       0  100.00%
R/LongitudinalRandomSlope.R       17       1  94.12%   43
R/LongitudinalSamples.R           15       0  100.00%
R/Parameter.R                     12       0  100.00%
R/ParameterList.R                 28       0  100.00%
R/Prior.R                         58       9  84.48%   127-135
R/simulations_gsf.R               43       0  100.00%
R/simulations_os.R                11       5  54.55%   35-39
R/simulations_rs.R                21       0  100.00%
R/simulations.R                   99       1  98.99%   122
R/StanModel.R                      7       0  100.00%
R/StanModule.R                   145       5  96.55%   185-186, 202, 235, 246
R/SurvivalExponential.R            9       0  100.00%
R/SurvivalLoglogistic.R           10      10  0.00%    31-40
R/SurvivalModel.R                 12       0  100.00%
R/SurvivalSamples.R              124       0  100.00%
R/SurvivalWeibullPH.R             10       0  100.00%
R/utilities.R                    146       1  99.32%   13
R/zzz.R                           11      11  0.00%    4-29
TOTAL                           1159      68  94.13%

Diff against main

Filename                 Stmts    Miss  Cover
---------------------  -------  ------  --------
R/DataJoint.R              +19       0  +100.00%
R/generics.R                -1       0  -0.76%
R/JointModel.R              -5      -3  +4.82%
R/JointModelSamples.R      -37       0  +100.00%
R/SurvivalSamples.R        +81       0  +100.00%
R/utilities.R              +30       0  +0.18%
R/zzz.R                     +1      +1  +100.00%
TOTAL                      +88      -2  +0.67%

Results for commit: 642e073

Minimum allowed coverage is 80%

♻️ This comment has been updated with latest results

[github-actions]

Unit Tests Summary

    1 files    25 suites   7m 13s ⏱️
  57 tests   56 ✔️ 1 💤 0 ❌
228 runs  227 ✔️ 1 💤 0 ❌

Results for commit 6c42723.

♻️ This comment has been updated with latest results.

[danielinteractive]

thx Craig, just a few minor things below

[gowerc]

Hey @danielinteractive , Think I addressed all your comments. I also changed the documentation for the SurvivalSamples to be a single page with the constructor (not sure why I hadn't done this). Appears to be an package version issue with the lintr which I've raised a query about with the IDR guys.