pharmpy.methods.frem.results module

class pharmpy.methods.frem.results.FREMResults(coefficients=None, parameter_variability=None, covariate_effects=None, individual_effects=None, unexplained_variability=None, covariate_statistics=None, covariate_effects_plot=None, individual_effects_plot=None, unexplained_variability_plot=None, covariate_baselines=None, parameter_inits_and_estimates=None, base_parameter_change=None, estimated_covariates=None, ofv=None)[source]

Bases: pharmpy.results.Results

FREM Results class

What follows is a description on how the FREM results are stored. See pharmpy.methods.frem for a description of how the results are calculated.

covariate_baselines

DataFrame with the baseline covariate values used in the analysis for each indiviual. Index is ID. One column for each covariate.

      WGT  APGR
ID
1.0   1.4   7.0
2.0   1.5   9.0
3.0   1.5   6.0
4.0   0.9   6.0
5.0   1.4   7.0
6.0   1.2   5.0
7.0   1.0   5.0
8.0   1.2   7.0
9.0   1.4   8.0
...   ...   ...
covariate_effects

DataFrame with covariate effects. Index is parameter, covariate and condition where condition can be either 5th or 95th. Columns are p5, mean and p95 for the 5th percentile of the effect, the estimated effect and the 95th percentile of the effect respectively. Effect sizes are in fractions. Example:

                                     p5      mean       p95
parameter covariate condition
ETA(1)    WGT       5th        0.901869  0.972920  1.051093
                    95th       0.903849  1.064605  1.233111
          APGR      5th        1.084238  1.248766  1.413923
                    95th       0.848297  0.901919  0.962312
ETA(2)    WGT       5th        0.942450  1.004400  1.068390
                    95th       0.874409  0.995763  1.127777
          APGR      5th        0.832008  0.924457  1.021307
                    95th       0.990035  1.039444  1.091288
covariate_effects_plot

Plot of the covariate effects generated by plot_covariate_effects()

covariate_statistics

DataFrame with summary statistics of the covariate baselines. Index is covariates. Columns are p5 with 95th percentile, mean, p95 with 95th percentile, stdev, ref with the used reference value, categorical and other with the non-reference value for categorical covariates. Example:

individual_effects

DataFrame with individual covariate effects. Index is ID and parameter. Columns are observed, p5 and p95 for the observed individual effect, 5th and 95th percentiles of the estimated individual effects respectively. Example:

                observed        p5       p95
ID   parameter
1.0  ETA(1)     0.973309  0.946282  0.982391
     ETA(2)     1.009411  0.995276  1.024539
2.0  ETA(1)     0.911492  0.832168  0.941760
     ETA(2)     1.036200  0.990081  1.091438
3.0  ETA(1)     1.013772  1.007555  1.028463
...                  ...       ...       ...
57.0 ETA(2)     0.987500  0.942709  1.031111
58.0 ETA(1)     0.939409  0.883782  0.956792
     ETA(2)     1.023321  0.993543  1.057408
59.0 ETA(1)     0.992578  0.952785  1.027261
     ETA(2)     0.999220  0.968931  1.033819
individual_effects_plot

Plot of the individual effects generated by plot_individual_effects()

unexplained_variability

DataFrame with remaining unexplained variability. Index is parameter and covariate. Covariate is none for no covariates and all for all covariates. Example:

                     sd_observed    sd_5th   sd_95th
parameter covariate
ETA(1)    none          0.195327  0.221382  0.298465
          WGT           0.194527  0.218385  0.292261
          APGR          0.182497  0.202192  0.279766
          all           0.178956  0.197282  0.268090
ETA(2)    none          0.158316  0.152510  0.210044
          WGT           0.158313  0.148041  0.207276
          APGR          0.155757  0.149037  0.203477
          all           0.155551  0.144767  0.201658
unexplained_variability_plot

Plot of the unexplained variability generated by plot_unexplained_variability()

parameter_inits_and_estimates

Initial parameter estimates and estimates after fitting for all intermediate models and the final model.

base_parameter_change

The relative change in parameter estimates from base model to the FREM model.

covariate_estimates

Model estimates of covariate statistics

parameter_variability

Conditioned parameter variability

coefficients

Parameter covariate coefficients. Calculated one at a time or all together.

add_plots()[source]

Create and add all plots to results object

plot_covariate_effects()[source]

Plot covariate effects

plot_individual_effects()[source]

Plot individual effects

plot_unexplained_variability()[source]

Plot unexplained variability

rst_path = PosixPath('/home/runner/work/pharmpy/pharmpy/pharmpy/.tox/docs/lib/python3.9/site-packages/pharmpy/methods/frem/report.rst')
pharmpy.methods.frem.results.add_base_vs_frem_model(res, frem_model, model_1)[source]
pharmpy.methods.frem.results.add_parameter_inits_and_estimates(res, frem_model, intermediate_models)[source]
pharmpy.methods.frem.results.calculate_results(frem_model, continuous, categorical, method=None, intermediate_models=None, seed=None, **kwargs)[source]

Calculate FREM results

Parameters

method – Either ‘cov_sampling’ or ‘bipp’

pharmpy.methods.frem.results.calculate_results_from_samples(frem_model, continuous, categorical, parvecs, rescale=True)[source]

Calculate the FREM results given samples of parameter estimates

pharmpy.methods.frem.results.calculate_results_using_bipp(frem_model, continuous, categorical, rescale=True, samples=2000, seed=None)[source]

Estimate a covariance matrix for the frem model using the BIPP method

Bootstrap on the individual parameter posteriors Only the individual estimates, individual unvertainties and the parameter estimates are needed.

pharmpy.methods.frem.results.calculate_results_using_cov_sampling(frem_model, continuous, categorical, cov_model=None, force_posdef_samples=500, force_posdef_covmatrix=False, samples=1000, rescale=True, seed=None)[source]

Calculate the FREM results using covariance matrix for uncertainty

Parameters

  • cov_model – Take the parameter uncertainty covariance matrix from this model instead of the frem model.

  • force_posdef_samples – The number of sampling tries before stopping to use rejection sampling and instead starting to shift values so that the frem matrix becomes positive definite. Set to 0 to always force positive definiteness.

  • force_posdef_covmatrix – Set to force the covariance matrix of the frem movdel or the cov model to be positive definite. Default is to raise in this case.

  • samples – The number of parameter vector samples to use.

pharmpy.methods.frem.results.get_params(frem_model, rvs, npars)[source]
pharmpy.methods.frem.results.psn_frem_results(path, force_posdef_covmatrix=False, force_posdef_samples=500, method=None)[source]

Create frem results from a PsN FREM run

Parameters

path – Path to PsN frem run directory

Returns

A FREMResults object

pharmpy.methods.frem.results.psn_reorder_base_model_inits(model, path)[source]

Reorder omega inits from base model in PsN

If base model was reordered PsN writes the omega inits dict to m1/model_1.inits

pharmpy.methods.frem.results.rename_duplicate(params, stem)[source]

Inheritance Diagram

Inheritance diagram of pharmpy.methods.frem.results

pharmpy.methods.frem.models module pharmpy.methods.linearize package