Fix computation of fraction time used by facility ID and officer type for clinic specific rescaling factors by sangeetabhatia03 · Pull Request #1773 · UCL/TLOmodel

sangeetabhatia03 · 2026-01-07T15:29:16Z

This PR fixes issue #1770.
In the implementation of clinics, we did not pass clinic information to the function frac_time_used_by_facID_and_officer, and therefore the computation therein was not clinic-specific, leading to erroneous calculations.
We have now fixed this and updated other necessary objects (namely running_total_footprint) to be clinic-specific.
This PR also brings in changes from PR #1662 and PR #1689 because in the absence of these, we would have to make the whole squeeze factor computation clinic-specific, when it is intended to be removed shortly.

… since we removal squeeze factors

…sed_on_squeeze_factors and when checking if mode 1 appointments can go ahead

…queeze' into sb/fix-1770

marghe-molaro

Hi @sangeetabhatia03, I think this is looking very good, I've just added a few comments to tighten the test a bit. I have also noticed that the HealthSystem seems to be behind the version in master, so it would be good to rebase just so that we're 100% sure we're not missing anything!

marghe-molaro · 2026-01-15T10:06:05Z

tests/test_healthsystem.py

+        s = sim.population.props[col]
+        ## Not specifying the dtype explicitly here made the col a string rather than a category
+        ## and that caused problems later on.
+        sim.population.props[col] = pd.Series(s.cat.categories[0], index=s.index, dtype=s.dtype)


The district of residence should be derived from the location associated with the selected Facility_ID, so that we can be sure the district to which the population is assigned matches the district of the facility

marghe-molaro · 2026-01-15T10:08:21Z

tests/test_healthsystem.py

+    # Now adjust capabilities available.
+    # GenericClinic has exactly the capability than needed to run all the appointments;
+    # Clinic1  has less capability than needed to run all the appointments;
+    # This will ensure rescaling factor for GenericClinic < 1


The rescaling factor can never be less than 1, we don't reduce capabilities if they are underused

marghe-molaro · 2026-01-15T10:08:57Z

tests/test_healthsystem.py

+    nevents_clinic1 = 90
+    sim = schedule_hsi_events(nevents_generic_clinic, nevents_clinic1, sim)
+
+    ## This hsi is only created to get the expected items; therefore the treatment_id is not important


if possible specify what these 'items' are, just so the intentions here are clearer

marghe-molaro · 2026-01-15T11:03:58Z

tests/test_healthsystem.py

+        sim.make_initial_population(n=tot_population)
+
+        sim.modules["HealthSystem"]._clinic_configuration = pd.DataFrame(
+            [{"Facility_ID": 20.0, "Officer_Type_Code": "DCSA", "Clinic1": 0.6, "GenericClinic": 0.4}]


Since this 0.6/0.4 split is arbitrary/never used, I would move the initialisation of the capabilities and clinics split here for facility 20 here (based on your intended plan of number of appts that you expect to be delivered in each).

marghe-molaro · 2026-01-15T11:05:53Z

tests/test_healthsystem.py

+
+        return sim
+
+    def schedule_hsi_events(notherclinic, nclinic1, sim):


I would try to be consistent with the use of 'other clinic' and 'generic clinic' if possible, for clarity

marghe-molaro · 2026-01-15T11:37:42Z

src/tlo/methods/healthsystem.py

@@ -2082,8 +1989,9 @@ def log_clinic_current_capabilities_and_usage(self, clinic_name):
        This will log the percentage of the current capabilities that is used at each Facility Type, according the


There is no need for log_clinic_current_capabilities_and_usage to be its own fnct as log_current_capabilities_and_usage is just calling that in a loop, only keep the latter to cut down on fncs a bit

marghe-molaro · 2026-01-15T14:41:46Z

src/tlo/methods/healthsystem.py

                _appt_footprint_before_running = event.EXPECTED_APPT_FOOTPRINT
-
-                # Mode 1: All HSI Events run with squeeze provided latter is not inf
+                # Mode 0: All HSI Event run, with no squeeze


Mode 0 should no longer be in master (removed by PR #1690), if you haven't rebased to master yet it would be good to do so, just so that we can make sure all changes are up to date

So I've just tried to do this and it appears that even after merging master the mode 0 functionality remains.
I think the issue is with the branching history of the sb/fix-1770 branch. @tamuri could you advise on how it'd best to solve this?

Seems like the branch for this PR did not originally branch off from master. One of the commits brought a lot of mode 0 back into this branch (by hand).

Unfortunately, no easy way to fix this. I would save a copy of the healthsystem.py in this branch. Then merge master and accept, entirely, the healthsystem.py in master. Then put back in the required changes (without bringing back in mode 0) from your saved copy.

tamuri · 2026-01-16T08:53:23Z

resources/healthsystem/human_resources/clinics/ResourceFile_ClinicConfigurations/Default.csv

-version https://git-lfs.github.com/spec/v1
-oid sha256:cd312903ff50d5233d81075b1f38e7879b8933e3ad7067d52c696e4f37e51eac
-size 44
+Facility_ID,Officer_Type_Code,Clinic1,GenericClinic


These two Default.csv resource files have not been committed using Git LFS. It might be that Git LFS is not setup correctly on your machine. You will need to: (1) git rm these two files (2) make sure Git LFS is setup properly on your machine (3) add the files back and then (4) commit.

tbhallett · 2026-01-20T18:27:21Z

Closing in favour of #1781

* Remove squeeze factor from HS * Ensure default logging for multiple treatment ids * Replaced rescaling of capabilities based on squeeze so that it is based on load * Ensure running footprint and load is clinic specific * Add clinics running footprint and test * For test to pass for now only log Capacity_By_FacID_and_Officer for GenericClinic * Style fixes * Fix test of squeeze logging * More logging fixes in test * Update comment in test because rescaling factor cannot be less than 1 * Consistent variable names * Remove redundant function log_current_capabilities_and_usage called log_clinic_current_capabilities_and_usage looping over the clinics. I have now removed log_clinic_current_capabilities_and_usage, and moved the looping logic to where the capabilities are logged. * Retrieve facility id and district from appointment type needs * Additional comments in tests to explain rationale I have updated comments in the test to explain the rationale for retaining an arbitrary clinic capabilities split in test set up. * WIP: looping over clinics to log * Debugging * All but one tests pass * WIP; debugging * Fix test rescaling capabilities based on load factors * Style fixes * Style fixes Signed-off-by: sangeetabhatia03 <sangeetabhatia03@gmail.com> * Remove print statements --------- Signed-off-by: sangeetabhatia03 <sangeetabhatia03@gmail.com> Co-authored-by: sangeetabhatia03 <sangeetabhatia03@gmail.com> Co-authored-by: Tim Hallett <39991060+tbhallett@users.noreply.github.com>

tamuri and others added 15 commits August 6, 2025 15:30

Remove squeeze factor logic

a761713

Remove unused import

b2aee5e

Remove hold-over of events when running mode 0 and 1 - there are none…

7e5d52c

… since we removal squeeze factors

Remove reference to squeeze factors in test_rescaling_capabilities_ba…

c0f9e88

…sed_on_squeeze_factors and when checking if mode 1 appointments can go ahead

Remove outdated test test_run_in_mode_1_with_almost_no_capacity

6bc4401

Style fixes

1606ff8

WIP; compute running footprint and rescaling factors by *clinic*

cccbd7b

Resolve merge conflicts

896066c

Merge remote-tracking branch 'origin/molaro/compute-load-instead-of-s…

fa7e0be

…queeze' into sb/fix-1770

WIP: working through the test suite

22c6f81

WIP: Amend test to loop over clinics while checking capacity logging

6ddd921

Run events in mode 0 or 1 with clinic-specific running footprint

6f14761

Fix key name in logger

466cc94

Check officer availability in clinic-specific bucket

9a8ed3b

Pull out clinic-specific rows in test

3eaedfa

sangeetabhatia03 requested a review from marghe-molaro January 7, 2026 15:29

sangeetabhatia03 added 10 commits January 7, 2026 15:49

Fix formatting errors

6adce88

Fix formatting

ecadc74

Pass clinic info when computing rescaling factor

fcbc46b

Record clinic in logger

56cd175

Assign inpatient bed-days to GenericClinic

d122c8f

Test rescaling factors for clinics

f4c9463

Resource files with 2 clinics

ebe3ff4

Linting

c9e2ad0

Map treament ids to clinic with non-zero capabilities

b1c313b

Missing else block

6b712f4

marghe-molaro mentioned this pull request Jan 14, 2026

Allow suspend/resume of simulations on Batch runs #1668

Merged

marghe-molaro reviewed Jan 15, 2026

View reviewed changes

tamuri reviewed Jan 16, 2026

View reviewed changes

marghe-molaro mentioned this pull request Jan 16, 2026

Fix branching issues in PR #1662, #1689, and #1773 #1781

Merged

tbhallett closed this Jan 20, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix computation of fraction time used by facility ID and officer type for clinic specific rescaling factors#1773

Fix computation of fraction time used by facility ID and officer type for clinic specific rescaling factors#1773
sangeetabhatia03 wants to merge 25 commits intomasterfrom
sb/fix-1770

sangeetabhatia03 commented Jan 7, 2026

Uh oh!

marghe-molaro left a comment

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026 •

edited

Loading

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026

Uh oh!

marghe-molaro Jan 15, 2026 •

edited

Loading

Uh oh!

tamuri Jan 16, 2026

Uh oh!

tamuri Jan 16, 2026

Uh oh!

tbhallett commented Jan 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants


		return sim

		def schedule_hsi_events(notherclinic, nclinic1, sim):

		@@ -2082,8 +1989,9 @@ def log_clinic_current_capabilities_and_usage(self, clinic_name):
		This will log the percentage of the current capabilities that is used at each Facility Type, according the

Conversation

sangeetabhatia03 commented Jan 7, 2026

Uh oh!

marghe-molaro 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!

marghe-molaro Jan 15, 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!

marghe-molaro Jan 15, 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!

tbhallett commented Jan 20, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

marghe-molaro Jan 15, 2026 •

edited

Loading

marghe-molaro Jan 15, 2026 •

edited

Loading