Last Mile LabA vaccine cold chain simulation · 120,000 doses, one month
Look
Graphics
Play along
Keys: + and − change depth, P opens the plan, Esc closes cards and menus.
21
Layers
1 / 5
Method
What is real, and what is simulated
Last Mile Lab is a seeded simulation of one month in a fictional province: 120,000 vaccine doses pass fourteen checks grouped in four gates. The population is synthetic. The shape of the losses is not: each rate sits inside a range reported in published cold chain studies.
Calibrated to
Freezing is common and invisible. Between 14% and 35% of refrigerators or transport shipments exposed vaccine to freezing temperatures (Matthias and others, systematic review, Vaccine, 2007). A later review found freezing still reported in storage and in transport, in richer and poorer countries alike (Hanson and others, Vaccine, 2017). In the model, about 10 in 100 doses freeze on average: fewer than the share of fridges, because not every exposure reaches every dose. Matthias 2007 · Hanson 2017
The vial monitor shows heat, not frost. Its inner square darkens with heat; it cannot show that a vaccine froze, which needs a separate freeze indicator or a shake test. VVM training module
Temperature, distribution and stock are the weak links. In WHO's Effective Vaccine Management assessments (2009 to 2020), temperature monitoring, vaccine distribution and information systems had medians below the 80% standard at every level, and only 16 countries reached 80% overall. That is why gates B and C carry most of the loss. WHO EVM global analysis
Big vials waste more. WHO's indicative maximum wastage for freeze-dried vaccines is 50% for 10 to 20 dose vials and 10% for 1 to 2 dose vials; the model's open-vial check follows that gradient. TechNet discussion quoting WHO
Only spoiled doses count. A dose that is not given stays in the fridge for another day, so the model counts a dose as lost only when it is damaged, freezes, overheats, expires or is thrown away from an opened vial. Children who are never reached are a coverage problem, outside this model: in 2024, 14.3 million infants received no vaccine at all and DTP3 coverage was 85%. WHO and UNICEF, July 2025
Synthetic
The province, its areas and facilities, the vials you can follow, the dose prices (illustrative, in the range of public UNICEF prices), the multipliers by area, session and vial size, and the monsoon and campaign effects. The lever effects are assumptions, not trial results. This is a teaching model, not a measurement of any country.
How to read it
Each ball is one dose from a fixed sample of 360; every figure covers the whole simulated month. Same seed, same month: Re-run draws the month again and figures move a little. The hourglass: the cube in the middle is the chain, the bulb above keeps what reaches a child, the bulb below collects what is lost.