WSI malaria prevention pilot v0.1
Evidence checked 9 October 2026

Reproduce the numerical results:
  python calculate.py source_inputs.json
This command uses only Python's standard library. Edit the profiles in
source_inputs.json to explore the USU mapping. Source facts are a frozen
extract of the linked GiveWell 2025 version 3 public model, Chad column.
Preserve source facts unless intentionally replacing the evidence snapshot.

Meaning of the result:
  Gross acute-malaria USU averted per USD 1,000 of all-contributor costs.
  3-59 month-olds in the Chad SMC program, 2026-27 grant scope.
  Under-5 incidence is the age proxy inherited from the source model.
  Already funded program benchmark, not marginal impact of a donation today.
  Fatal pre-death courses replace nonfatal courses. Death itself has no USU.
  Direct malaria deaths are displayed separately. Indirect deaths excluded.
  Side effects, long-term sequelae and other excluded effects are unquantified.
  Mapping scenarios are judgmental sensitivity cases, not confidence limits.

Files:
  source_inputs.json: Exact model inputs, source cells, URLs and WSI profiles.
  calculate.py: Portable independent numerical reproduction.
  results.json: Checked results from the delivered workbook's build.
  build.mjs: Workbook authoring source. Requires @oai/artifact-tool. Place this
    beside source_inputs.json. WSI_OUTPUT_DIR and WSI_QA_DIR can set output
    locations. This builder uses the fixed v0.1 profile defaults in its code;
    use the workbook or calculate.py for profile exploration.
  build_note.py: PDF authoring source. Requires reportlab and DejaVu fonts.
    Set WSI_OUTPUT_DIR to the folder containing results.json and inputs.

Validation:
  Workbook calculations match an independent calculation and the published
  model's episode total and direct share of deaths, on the same cost basis.
  Tested zero budget, burden multiplier, cost multiplier, intensity changes,
  and missing profile/budget inputs. All restored before final export.
  No formula errors found. PDF and worksheet renderings inspected.
  Not tested in desktop Excel.

Refinement priorities:
  1. Better symptom duration and experienced intensity for the target children.
  2. Updated local malaria burden, including independent mortality sensitivity.
  3. Quantify intervention side effects before claiming net suffering averted.
  4. Match a current unfunded campaign to its costs and additional benefit.
  5. Reconcile the public narrative's $8.80 with the linked CEA's cost values.
