Vivax competing hazards main drugs #330

RJSheppard · 2024-07-30T13:16:02Z

Adding drug vivax parameters, including liver stage prophylaxis.

pwinskill · 2024-08-01T12:58:58Z

R/drug_parameters.R

 #' @export
 DHA_PQP_params <- c(.95, 0.09434, 4.4, 28.1)

 #' @title Preset parameters for the AL drug
 #' @description From SI of Commun. 5:5606 doi: 10.1038/ncomms6606 (2014)
 #' @details Use a vector of preset parameters for the AL drug (artemether-lumefantrine)
 #' @details Default parameters, from L to R, are: drug_efficacy: 0.95, drug_rel_c: 0.05094, drug_prophylaxis_shape: 11.3, drug_prophylaxis_scale: 10.6
+#' @export
+AL_params_falciparum <- c(.95, 0.05094, 11.3, 10.6)
+
 #' @export
 AL_params <- c(.95, 0.05094, 11.3, 10.6)


Is there a reason to keep the non-species specific drugs? Backward compatability?

I would lean towards keeping AL_params and removing AL_params_falciparum. Bumping the version and getting everyone to change their code is not convenient

pwinskill · 2024-08-01T13:08:36Z

R/human_infection.R

@@ -370,9 +370,28 @@ relapse_bite_infection_hazard_resolution <- function(
  # get bite infections
  bite_infections <- infected_humans$copy()$and(relapse_infections$not(inplace = F))

+  ## drug prophylaxis may limit formation of new hypnozoite batches
+  ls_prophylaxis <- rep(0, bite_infections$size())
+  if(length(parameters$drug_hypnozoite_efficacy)>0){


Finding this section a little hard to follow.

In set_drugs() or set_clinical_treatment() I wasn't able to see any checks that if species == vivax, all drugs have the additional hypnozoite elements. If it isn't in there it might be good to have a check. What happens if these are NA in the simulation?

For example if I set drugs as:

drugs <- list( malariasimulation::AL_params, malariasimulation::CQ_PQ_params_vivax )

In my parameters I get parameters$drug_hypnozoite_efficacy = NA 0.713, So length(parameters$drug_hypnozoite_efficacy)>0isTRUEhere. I think this is getting dealt with in line 379ls_medicated[ls_drug > 0] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])`, but I wonder if it should be checked and raised as an issue to the user when setting parameters. That would be more transparent?

pwinskill · 2024-08-01T13:08:54Z

R/human_infection.R

  ## all bitten humans with an infectious bite (incorporating prophylaxis) get a new batch of hypnozoites
  if(bite_infections$size()>0){
-    new_hypnozoite_batch_formed <- bite_infections # bitset_at(bite_infections, bernoulli_multi_p(1-ls_prophylaxis))
+    new_hypnozoite_batch_formed <- bitset_at(bite_infections, bernoulli_multi_p(1-ls_prophylaxis))


Suggested change

new_hypnozoite_batch_formed <- bitset_at(bite_infections, bernoulli_multi_p(1-ls_prophylaxis))

new_hypnozoite_batch_formed <- bitset_at(bite_infections, bernoulli_multi_p(1 - ls_prophylaxis))

pwinskill · 2024-08-01T13:10:04Z

R/mda_processes.R

@@ -76,6 +79,19 @@ create_mda_listeners <- function(
      variables$drug$queue_update(drug, to_move)
      variables$drug_time$queue_update(timestep, to_move)
    }
+
+    # Update liver stage drug effects
+    if(!is.na(parameters$drug_hypnozoite_efficacy[drug])){


Same comment as above. For clarity should all vivax drugs have this parameter (even if by default the user sets it to 0)?

…ers. Default drug parameters have been relabelled using parasite species names, although the original names have been left in for now, so that it can still be compatable with previous code. All the tests and vignettes use the parasite named parameters sets. Treatment, MDA and PMC code now includes hypnozoite clearance when these drugs are used. Added drug prophylaxis effects to human infection, preventing new hypnozoite batch formation.

RJSheppard · 2024-10-25T10:57:12Z

/benchmark

github-actions · 2024-10-25T12:02:31Z

This is how benchmark results would change (along with a 95% confidence interval in relative change) if a24ebdd is merged into dev:

✔️large_population: 2.37m -> 2.33m [-3.82%, +0.47%]
✔️small_population: 31.8s -> 32.1s [-1.12%, +2.87%]

Further explanation regarding interpretation and methodology can be found in the documentation.
Plots and raw data are available as artifacts of the workflow run.

pwinskill

Thanks @RJSheppard.

A few Qs:

I'm still a bit unclear as to why we have, for exampe both SP_AQ_params_falciparum and SP_AQ_params. When is the second used?
I see we can use vivax drugs for MDA. What have we done with SMC and PMC? To my knowledge those aren't avaiable for vivax-specific implementation. Is it worth putting in a check to error if users try to set up either of those interventions with vviax drugs? I can't remeber but perhaps we already have a check to stop those interventions completely if parasite = vivax?
Worth sneeking in a fix for inconsistency in set_drug vector lengths does not introduce error #339 here?

pwinskill · 2024-10-28T08:41:31Z

R/human_infection.R

@@ -729,6 +749,15 @@ calculate_treated <- function(
    )
  }

+  # Update liver stage drug effects
+  if(length(parameters$drug_hynozoite_efficacy) > 0){


Are both ifs required here?
If successfully_treated$successfully_treated_hypnozoites$size() > 0 wouldn't length(parameters$drug_hynozoite_efficacy) > 0 by definition?

The way I've coded it means that successfully_treated$successfully_treated_hypnozoites is only generated when length(parameters$drug_hynozoite_efficacy) > 0. So it doesn't exist for falciparum, or for vivax when non hypnozoite drugs are being used (e.g., just CQ).

This might be avoidable if all vivax drugs have that parameter (see other comment)?

giovannic · 2024-10-28T11:41:37Z

R/drug_parameters.R

+#' @details Default parameters, from L to R, are: drug_efficacy: 1, drug_rel_c: 0.5, drug_prophylaxis_shape: 5, drug_prophylaxis_scale: 20,
+#' drug_hypnozoite_efficacy: 0.713, drug_hypnozoite_prophylaxis_shape: 5.5, drug_hypnozoite_prophylaxis_scale: 30
+CQ_TQ_params_vivax <- c(1, 0.5, 5, 20, 0.713, 5, 30)
+
 #' @title Parameterise drugs to use in the model
 #'
 #' @param parameters the model parameters
 #' @param drugs a list of drug parameters, can be set using presets
 #' @export
 set_drugs <- function(parameters, drugs) {


Is there any validation we could do here to validate if parameters$parasite == vivax that the correct drugs are used? Or do all drugs make sense for all parasites?

I spoke with Pete about this a while ago, and we came to a place of hoping that people would use their good sense to choose which parameter sets to apply for each species. While drug parameters are parasite-specific, there might be circumstances where someone would want to see what happens if the same drug is applied to both species. I've added a few warnings to flag where people might do something strange while maintaining flexibility. I could change some of the warnings to errors (e.g., the length of the parameter inputs) if we think that's important.

giovannic · 2024-10-28T11:46:29Z

R/human_infection.R

+
+      ls_drug <- variables$ls_drug$get_values(bite_infections)
+      ls_medicated <- (ls_drug > 0)
+      ls_medicated[ls_drug > 0] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])


I would expect this to fail? Since length(ls_medicated[ls_drug > 0]) is less than or equal to length(ls_drug). If some people are not medicated then there'll be a weird assignment here. Is this tested for?

Suggested change

ls_medicated[ls_drug > 0] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])

ls_medicated <- ls_medicated & !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])

I see what you're saying. An issue is that some of ls_drug will be 0 (if no drugs have been assigned to the individual), which means that !is.na(parameters$drug_hypnozoite_efficacy[ls_drug]) may not have the same length as ls_medicated (as far as I can work it out). I've taken a look to make sure it's doing what it's supposed to do, with a few tests to check it through.

giovannic · 2024-10-28T11:55:53Z

R/drug_parameters.R

 #' @export
 DHA_PQP_params <- c(.95, 0.09434, 4.4, 28.1)

 #' @title Preset parameters for the AL drug
 #' @description From SI of Commun. 5:5606 doi: 10.1038/ncomms6606 (2014)
 #' @details Use a vector of preset parameters for the AL drug (artemether-lumefantrine)
 #' @details Default parameters, from L to R, are: drug_efficacy: 0.95, drug_rel_c: 0.05094, drug_prophylaxis_shape: 11.3, drug_prophylaxis_scale: 10.6
+#' @export
+AL_params_falciparum <- c(.95, 0.05094, 11.3, 10.6)
+
 #' @export
 AL_params <- c(.95, 0.05094, 11.3, 10.6)


I would lean towards keeping AL_params and removing AL_params_falciparum. Bumping the version and getting everyone to change their code is not convenient

RJSheppard · 2024-10-31T17:12:43Z

I've removed the _falciparum drugs names for backwards compatibility.

vivax drugs can currently be used in SMC/PMC via the update_mass_drug_admin function that is primarily called during MDA (see mda_processes.R). It may not be realistic for this to be used in those settings, but for consistency/flexibility I've left it in. I can imagine circumstances where it might be interesting to look at TQ prophylaxis as SMC/PMC, given it's longer half-life.

I can take a look at the drug vector length issue, but the example code is only causing a problem because the set_drugs function hasn't been used. The parameters are changed manually in the example, which isn't advised. I've added some warnings in set_drugs for when the vector length doesn't match up (and could make these errors if we want to fully restrict people from adding incorrect drug vector lengths).

… Warnings added to set_drugs function for correct parasite drug parameter attribution. Tests to make sure liver stage prophylaxis is correct.

…vignette to reference drug impacts and prophylaxis.

giovannic

Looks good thanks!

I've commented in line on a potentially redundant line, let me know if I'm mistaken.

giovannic · 2024-11-11T13:50:01Z

R/human_infection.R

+
+    ls_drug <- variables$ls_drug$get_values(bite_infections)
+    ls_medicated <- ls_drug > 0
+    ls_medicated[ls_medicated] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])


This is quite confusing. And possibly redundant? Don't we expect all liver stage drugs have a hypnozoite efficacy?

If I understand the question, the way I've coded it is that the parameter sets are parasite specific, rather than being blood or liver stage specific. That means that vivax drug parameter sets can be either only blood stage (where they would have 4 parameters), or would have both blood and liver stage parameters (which total 7 parameters), and not all "vivax drugs" have liver stage action. Does that answer the question? (Happy to change things if you have a better way of doing things!)

Would it be more consistent to have liver-stage parameters for each vivax drug, and just set them to 0 if the drug is assumed to have no liver-stage impact?

…ponding aspect of vivax drug test.

pwinskill

Thanks @RJSheppard - just a few minor followups from me

pwinskill · 2024-11-18T14:55:00Z

R/human_infection.R

+
+    ls_drug <- variables$ls_drug$get_values(bite_infections)
+    ls_medicated <- ls_drug > 0
+    ls_medicated[ls_medicated] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])


Would it be more consistent to have liver-stage parameters for each vivax drug, and just set them to 0 if the drug is assumed to have no liver-stage impact?

pwinskill · 2024-11-18T14:57:15Z

R/human_infection.R

@@ -729,6 +749,15 @@ calculate_treated <- function(
    )
  }

+  # Update liver stage drug effects
+  if(length(parameters$drug_hynozoite_efficacy) > 0){


This might be avoidable if all vivax drugs have that parameter (see other comment)?

RJSheppard requested review from giovannic and pwinskill July 30, 2024 13:16

pwinskill requested changes Aug 1, 2024

View reviewed changes

RJSheppard force-pushed the vivax_competing_hazards_main_equilibrium branch 2 times, most recently from b807634 to 16cc8c4 Compare September 18, 2024 14:56

RJSheppard force-pushed the vivax_competing_hazards_main_drugs branch from a34e336 to bbb1ad4 Compare September 18, 2024 19:56

RJSheppard force-pushed the vivax_competing_hazards_main_equilibrium branch from 16cc8c4 to a75a671 Compare September 19, 2024 13:00

RJSheppard force-pushed the vivax_competing_hazards_main_drugs branch from bbb1ad4 to f5a50c4 Compare September 19, 2024 13:18

RJSheppard force-pushed the vivax_competing_hazards_main_equilibrium branch from a75a671 to 2e0e3f3 Compare September 19, 2024 14:56

RJSheppard force-pushed the vivax_competing_hazards_main_drugs branch from f5a50c4 to d5fa60e Compare September 19, 2024 15:34

RJSheppard force-pushed the vivax_competing_hazards_main_equilibrium branch from 2e0e3f3 to 6309897 Compare September 20, 2024 16:29

RJSheppard force-pushed the vivax_competing_hazards_main_drugs branch from d5fa60e to 0582fec Compare September 20, 2024 16:38

RJSheppard force-pushed the vivax_competing_hazards_main_equilibrium branch 2 times, most recently from 921c84f to 16666ba Compare October 22, 2024 16:17

Base automatically changed from vivax_competing_hazards_main_equilibrium to dev October 25, 2024 10:14

RJSheppard force-pushed the vivax_competing_hazards_main_drugs branch from 0582fec to a24ebdd Compare October 25, 2024 10:56

pwinskill reviewed Oct 28, 2024

View reviewed changes

giovannic requested changes Oct 28, 2024

View reviewed changes

RJSheppard added 2 commits October 31, 2024 19:18

Drug names for falciparum are consistent for backwards compatability.…

8ebc04e

… Warnings added to set_drugs function for correct parasite drug parameter attribution. Tests to make sure liver stage prophylaxis is correct.

Added CQ_TQ_params_vivax to exported objects. Added some text to the …

9ad598f

…vignette to reference drug impacts and prophylaxis.

giovannic approved these changes Nov 11, 2024

View reviewed changes

Removed unnecessary code from ls prophylaxis calculation, with corres…

8930897

…ponding aspect of vivax drug test.

pwinskill reviewed Nov 18, 2024

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Vivax competing hazards main drugs #330

Vivax competing hazards main drugs #330

RJSheppard commented Jul 30, 2024

pwinskill Aug 1, 2024

giovannic Oct 28, 2024

pwinskill Aug 1, 2024

pwinskill Aug 1, 2024

pwinskill Aug 1, 2024

RJSheppard commented Oct 25, 2024

github-actions bot commented Oct 25, 2024

pwinskill left a comment

pwinskill Oct 28, 2024

RJSheppard Oct 31, 2024

pwinskill Nov 18, 2024

giovannic Oct 28, 2024

RJSheppard Oct 31, 2024

giovannic Oct 28, 2024

RJSheppard Oct 31, 2024

giovannic Oct 28, 2024

RJSheppard commented Oct 31, 2024

giovannic left a comment

giovannic Nov 11, 2024

RJSheppard Nov 11, 2024

pwinskill Nov 18, 2024

pwinskill left a comment

pwinskill Nov 18, 2024

pwinskill Nov 18, 2024

	new_hypnozoite_batch_formed <- bitset_at(bite_infections, bernoulli_multi_p(1-ls_prophylaxis))
	new_hypnozoite_batch_formed <- bitset_at(bite_infections, bernoulli_multi_p(1 - ls_prophylaxis))

	ls_medicated[ls_drug > 0] <- !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])
	ls_medicated <- ls_medicated & !is.na(parameters$drug_hypnozoite_efficacy[ls_drug])

Vivax competing hazards main drugs #330

Are you sure you want to change the base?

Vivax competing hazards main drugs #330

Conversation

RJSheppard commented Jul 30, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

RJSheppard commented Oct 25, 2024

github-actions bot commented Oct 25, 2024

pwinskill left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

RJSheppard commented Oct 31, 2024

giovannic left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

pwinskill left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment