Param Revamp - Contraception

[mmsuarezcosta]

Parameters that remain hardcoded

• Dateoffsets
• consumable and dosages

Files that I changed:
- Discontinuation_ByAge
- Initiation_byage
- Pregnancy_NotUsing_HIVeffect
- Pregnancy_NotUsing_In_2010
- Simplified_labour_parameters
- ResourceFile_ContraceptionAnalysisParams.csv
- ResourceFile_ContraceptionParams.csv

Files that require refactoring in order to properly label:
- Discontinuation_ByMethod
- Failure_ByMethod
- Initiation_AfterBirth
- Initiation_ByMethod
- Interventions_Pop
- Interventions_PPFP
- Method_Use_In_2010
- Prob_Switch_From
- Prob_Switch_From_And_To

Parameters requiring revision

• Local: Pregnancy_NotUsing_In_2010, Interventions_Pop, Interventions_PPFP
• Universal:
  - Pregnancy_NotUsing_HIVeffect (relative rate), Failure_ByMethod, rr_fail_under25 (not sure if this one takes into account the adherence rates of individuals in specific countries or if it is purely based on the efficacy of the contracpetion)
  - min_age_contraception and max_age_contraception : These is based off of the Guttmacher paper which is 'universal', so maybe these ranges should be defined as universal?
• Undetermined: Assume that all variables leveraged in the functions time_age_trend_in_initiation and time_age_trend_in_stopping were created for calibration purposes and thus are labeled as such

Recommendations:

• Hard code parameters from ResourceFile_ContraceptionAnalysisParams.csv in the analysis file to minimize the number of additional resource files
• Restructure contraception folder to same structure as other (i.e. rename 'contraception' folder to ResourceFile_Contraception and place all files here)

[mmsuarezcosta]

Ready for @marghe-molaro and @mnjowe review

[mnjowe]

Thanks Mariana @mmsuarezcosta for the great work. This is looking great!

My comments are as follows:

For resource files Initiation_ByAge.csv and Discontinuation_ByAge.csv

• Please make sure you are assigning correct prio_min for the negative values

For the resourcefile ResourceFile_ContraceptionParams.csv;

• Are the values for age_modification_factors parameter probabilities or rates?

• We are running the simulation from 2010 - 2099 so I don't know if the below is the correct assignment of prior min and prior max.

      parameter_name               value           prior_min   prior_max
      min_simulation_year         2010,                 2000        2020
      max_simulation_year.        2101,                 2050        2150

[mnjowe]

the reason why I have left a comment on why we should reconsider the values for both prior_min and prior_max on the years

[mmsuarezcosta]

My understanding is that the vague prior is helpful in case we want to change from running it from its current values, what other ranges can we consider

[mmsuarezcosta]

2010

Thanks for your review @mnjowe !

• age_modification_factors: I have interpreted these as rates
• min_simulation_year, max_simulation_year: For the priors, I have left them very vague as to allow for modelers to determine if they would like to change the date ranges. Given that our simulations never start before 2010, I will change the prior_min of min_simulation_year to 2010. For the other, I have left it wide based on how far into the future developers would like to simulate. Please recommend a different range if you believe we should update. Thanks.

[marghe-molaro]

Hi @mmsuarezcosta and @mnjowe,

min_simulation_year and max_simulation_year are scenario parameters, not undetermined, as they just describe the set up of the runs, so by definition they don't have a prior.

Other than this, I am happy for @EvaJanouskova as module leader to review. Eva, it would be particularly helpful if you confirm what the files used for analysis are, as we would like to move these to a different location. Many thanks in advance!

[mmsuarezcosta]

Hi @mmsuarezcosta and @mnjowe,

min_simulation_year and max_simulation_year are scenario parameters, not undetermined, as they just describe the set up of the runs, so by definition they don't have a prior.

Other than this, I am happy for @EvaJanouskova as module leader to review. Eva, it would be particularly helpful if you confirm what the files used for analysis are, as we would like to move these to a different location. Many thanks in advance!

@marghe-molaro Makes sense, I have updated the parameter CSV. Thanks!

[marghe-molaro]

@tbhallett ready for your review

[tbhallett]

Checking calibration looks OK: https://github.com/UCL/TLOmodel/actions/runs/19763622530

[tbhallett]

Hi @mnjowe --- please could you re-review this and if you're happy Approve?

[mnjowe]

I saw one test failing in the previous commit. Approving this hopping @tbhallett you will look into it