HaConfig: added art_initiation_rate by CarelPretorius · Pull Request #356 · hivtools/leapfrog

CarelPretorius · 2026-07-09T08:35:34Z

Adding functionality to calc ART initiation directly from a input rate

HaConfig: art_initiation_rate
adult_sim: added code to use art initiation rate to calc new art initiation
goals_sim: : added code to use art initiation rate to calc new art initiation
HvConfig: renamed an input parameter

adult_sim: added code to use art initiation rate to calc new art initiation goals_sim: : added code to use art initiation rate to calc new art initiation HvConfig: renamed an input parameter

CarelPretorius · 2026-07-09T08:36:09Z

               if constexpr (ModelVariant::run_goals) {
                   if(t > pars.hv.goals_base_year_idx){ 
-                    temp_art_adult_dropout = -std::log(1.0 - pars.hv.art_interupt_rate(t) *
+                    temp_art_adult_dropout = -std::log(1.0 - pars.hv.art_interrupt_rate(t) *


changed the name tpo: art_interrupt_rate

Is the conversion of the interruption rate from a percent to a per-capita rate via the log call needed here? Leapfrog might already handle that as a preprocessing step.

CarelPretorius · 2026-07-09T08:37:23Z

-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 
+      if (pars.hv.art_cov_num_percent[0] == ART_INIT_RATE) {
+      real_type NewInf = 0.0;
+      real_type PLHIV = 0.0;
+      real_type OnART = 0.0;
+      for (int ha = 0; ha < hAG; ++ha) {
+        const int a = ha + SS::p_idx_hiv_first_adult;
+        NewInf +=  c_ha.p_infections(a, s);       
+        PLHIV += c_ha.p_hivpop(a, s);             
+        for (int hm = i_ha.everARTelig_idx; hm < hDS; ++hm) {
+            for (int hu = 0; hu < hTS; ++hu) {
+              OnART += c_ha.h_artpop(hu, hm, ha, s);
+            }
+        }
+      }
+
+      i_ha.artinit_hts = std::max( opts.dt * pars.ha.art_initiation_rate(s,t) * (NewInf + PLHIV - OnART), 0.0);
+    }
+    else{


calculating ART initiation rate directly from input. The previous calc of new ART is in the second block of the iff statement. if (pars.hv.art_cov_num_percent[0] == ART_INIT_RATE) . A SS variable can be created for int ART_INIT_RATE = 5;

CarelPretorius · 2026-07-09T08:37:55Z

+    },
+
+    "art_initiation_rate": {
+      "num_type": "real_type",
+      "dims": ["SS::NS", "opts.proj_steps"]


added this input var

CarelPretorius · 2026-07-09T08:38:08Z

    },

-    "art_interupt_rate": {
+    "art_interrupt_rate": {


corrected the var name

CarelPretorius · 2026-07-09T08:39:01Z

+
+          if( pars.hv.art_cov_num_percent[0] == ART_INIT_RATE ){
+
+            real_type NewInf = c_hv.new_inf_vrs(v, rg, s); 
+            real_type PLHIV = 0.0;
+            real_type OnART = 0.0;
+            for (int hd = CD4_GT500; hd <= CD4_LT50; ++hd) {
+              PLHIV += c_hv.adults(v, rg, hd, s);
+              OnART += c_hv.adults(v, rg, hd + hOnArt, s);
+            }
+
+            start_art[v][rg][s] = std::max( opts.dt * pars.ha.art_initiation_rate(s,t) * (NewInf + PLHIV - OnART), 0.0);


calculating new ART directly from an input rate, instead of from a coverage

Do we actually need to tally then deduct OnART here? It looks like PLHIV might already include just people not on ART.

r-ash · 2026-07-09T10:16:59Z

    for (int s = 0; s < NS; ++s) {
-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 


You can make this static constexpr, this will ensure that ART_INIT_RATE is created at compile time and always refers to the same variable (no allocation within the loop)

Perhaps we should have a chat about this though, is there any reason we keep this as value 5? We have 3 cases?

Passed in art cov as number

Passed in art cov as %

Passed in art initiation rate

Can we mix between art initiation rate per year?

Perhaps we could also use a real enum here

addressed in slack message: should work for now but a more general implementation can follow

Using ART_INIT_RATE = 5 is an artifact from the Delphi implementation that exists because of the various ways adult ART uptake can be specified in Spectrum and the chronological order in which those were added to the software. It would be best to hide or abstract these artifacts from Leapfrog, but I defer to folks who know Leapfrog better what approach makes the most sense.

r-ash · 2026-07-09T10:19:05Z

+      real_type NewInf = 0.0;
+      real_type PLHIV = 0.0;
+      real_type OnART = 0.0;
+      for (int ha = 0; ha < hAG; ++ha) {
+        const int a = ha + SS::p_idx_hiv_first_adult;
+        NewInf +=  c_ha.p_infections(a, s);       
+        PLHIV += c_ha.p_hivpop(a, s);             
+        for (int hm = i_ha.everARTelig_idx; hm < hDS; ++hm) {
+            for (int hu = 0; hu < hTS; ++hu) {
+              OnART += c_ha.h_artpop(hu, hm, ha, s);
+            }
+        }


Can we fix the indentation here please.

r-ash · 2026-07-09T10:21:59Z

-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 
+      if (pars.hv.art_cov_num_percent[0] == ART_INIT_RATE) {


We should change this name, it is misleading. In fact this won't work to refer to some data from the goals model here in the adult model sim.

We pass this in with length 5 but then only ever use the 0th index of it? What is this input?

It has 5 potential values all in [0] since it comes from a radio button control group. This works for now but we can also change the input for to make it less confusing

happy to change the name pars.hv.art_cov_num_percent

I agree with Rob here, the art_cov_num_percent name didn't anticipate all the various ways we can specify ART coverage, so it seems best to make the name more general, e.g., art_entry_option.

r-ash · 2026-07-09T10:23:00Z

+      for (int ha = 0; ha < hAG; ++ha) {

-          const int a = ha + p_idx_hiv_first_adult;
+          const int a = ha + SS::p_idx_hiv_first_adult;
          //adults, plhiv not on art


Was this an intended change?

r-ash

This function is getting a bit big, I wonder if we try and refactor it to make it a bit easier to read, extract the if/else block into 2 helper functions computeArtInitFromRate(..) and computeArtInitFromCoverage(..) or similar?

r-ash · 2026-07-09T10:44:30Z

-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 
+      if (pars.hv.art_cov_num_percent[0] == ART_INIT_RATE) {


We pass this in with length 5 but then only ever use the 0th index of it? What is this input?

r-ash · 2026-07-09T10:48:15Z

    for (int s = 0; s < NS; ++s) {
-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 


Perhaps we could also use a real enum here

rlglaubius

I've left a few comments and questions on some of the open discussions here.

rlglaubius · 2026-07-10T14:29:38Z

+
+          if( pars.hv.art_cov_num_percent[0] == ART_INIT_RATE ){
+
+            real_type NewInf = c_hv.new_inf_vrs(v, rg, s); 
+            real_type PLHIV = 0.0;
+            real_type OnART = 0.0;
+            for (int hd = CD4_GT500; hd <= CD4_LT50; ++hd) {
+              PLHIV += c_hv.adults(v, rg, hd, s);
+              OnART += c_hv.adults(v, rg, hd + hOnArt, s);
+            }
+
+            start_art[v][rg][s] = std::max( opts.dt * pars.ha.art_initiation_rate(s,t) * (NewInf + PLHIV - OnART), 0.0);


Do we actually need to tally then deduct OnART here? It looks like PLHIV might already include just people not on ART.

rlglaubius · 2026-07-10T14:31:38Z

               if constexpr (ModelVariant::run_goals) {
                   if(t > pars.hv.goals_base_year_idx){ 
-                    temp_art_adult_dropout = -std::log(1.0 - pars.hv.art_interupt_rate(t) *
+                    temp_art_adult_dropout = -std::log(1.0 - pars.hv.art_interrupt_rate(t) *


Is the conversion of the interruption rate from a percent to a per-capita rate via the log call needed here? Leapfrog might already handle that as a preprocessing step.

rlglaubius · 2026-07-10T14:49:35Z

    for (int s = 0; s < NS; ++s) {
-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 


Using ART_INIT_RATE = 5 is an artifact from the Delphi implementation that exists because of the various ways adult ART uptake can be specified in Spectrum and the chronological order in which those were added to the software. It would be best to hide or abstract these artifacts from Leapfrog, but I defer to folks who know Leapfrog better what approach makes the most sense.

rlglaubius · 2026-07-10T15:00:14Z

-      i_ha.Xart_15plus = 0.0;
+
+      int ART_INIT_RATE = 5; 
+      if (pars.hv.art_cov_num_percent[0] == ART_INIT_RATE) {


I agree with Rob here, the art_cov_num_percent name didn't anticipate all the various ways we can specify ART coverage, so it seems best to make the name more general, e.g., art_entry_option.

HaConfig: added art_initiation_rate

02cb23d

adult_sim: added code to use art initiation rate to calc new art initiation goals_sim: : added code to use art initiation rate to calc new art initiation HvConfig: renamed an input parameter

CarelPretorius commented Jul 9, 2026

View reviewed changes

r-ash self-requested a review July 9, 2026 10:10

r-ash requested changes Jul 9, 2026

View reviewed changes

r-ash requested review from jeffeaton and rlglaubius July 10, 2026 14:05

rlglaubius reviewed Jul 10, 2026

View reviewed changes

Uh oh!

Conversation

CarelPretorius commented Jul 9, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

r-ash Jul 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

r-ash left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

rlglaubius left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

r-ash Jul 9, 2026 •

edited

Loading