Inspiration
A clinic power outage raises two distinct questions. How much cumulative heat exposure has each vaccine lot absorbed, and did any freeze-sensitive lot spend a full continuous hour below the freeze-alarm threshold? Those two questions often have different answers. ColdChainGuard turns that distinction into an interactive, inspectable offline rehearsal workbench using strictly fictional lots.
What it does
- WHO PQS VVM Arrhenius and Freeze-Excursion Engine. Models cumulative time-temperature exposure calibrated to published WHO PQS VVM2, VVM7, VVM14, and VVM30 endpoint pairs while running an independent continuous-duration clock for freeze-sensitive alum-adjuvant alarms.
- Lumped-Capacitance Thermal and PCM Outage Simulator. Rehearse a 48-hour monsoon solar-fridge outage, a 42 C heat-dome blackout, or a busy outreach carrier. Compare an unmanaged baseline against a plan that batches door openings, conditions phase-change thermal packs, and evaluates available staffed transfer windows.
- Bilingual 1-Page Action Sheet and Auditable Exports. Exports a one-page English or Spanish printable clinic checklist, an RFC 5545 UTC check-in .ics file, formula-neutralized CSV excursion logs, a canonical SHA-256 JSON audit bundle with deterministic replay, and a zero-backend AES-256-GCM Web Crypto fragment-sealed handoff summary.
How we built it
ColdChainGuard is built with native ES-module JavaScript, HTML5, CSS3, inline SVG, Web Crypto, and a same-origin Service Worker with zero runtime dependencies and zero external requests. A heat-balance ODE tracks chamber and PCM pack temperatures across latent-heat plateaus, Gauss-Legendre integration evaluates Arrhenius heat exposure, and all exports run locally in the browser.
Evaluation
The physical engine was verified against published WHO PQS calibration pairs and an analytic heat-transfer trajectory. All 397 automated Node tests passed, covering energy conservation, phase-change clamping, exact freeze-alarm timing, seeded input sweeps, and tamper rejection on encrypted handoffs and SHA-256 audit bundles. Headless Chrome verified 1280x800 desktop and 390x844 mobile layouts with zero console errors, zero external requests, offline reload, and one-page English and Spanish printouts.
- Monsoon outage (synthetic). Modeled doses on hold drop from 120 to 0 of 560, and time to first 2 to 8 C band exit extends from 13.0 hours to 72.0 hours.
- Heat dome blackout (synthetic). Modeled doses on hold drop from 200 to 120 of 560, while first band exit shifts from 9.0 to 2.4 hours. The workbench explicitly flags that a modeled hold still occurs near hour 12.6 before the first staffed transfer at hour 16, requiring earlier emergency escalation.
- Outreach carrier (synthetic). Modeled doses on hold drop from 300 to 0 of 560, freeze alarms drop from 1 to 0, and time to first band exit extends from 0.05 to 72.0 hours.
These are deterministic educational results on fictional scenarios, not clinical outcomes or proof of vaccine potency. ColdChainGuard is not a medical device or clinical release authorization. For real vaccines, follow physical VVM color checks, calibrated data loggers, and national immunization guidelines.
Challenges we ran into
Ranking plans strictly by doses avoiding both heat and freeze triggers showed how a safer plan can leave the 2 to 8 C window earlier while still protecting far more lots. Hardening Gauss-Legendre integration across steep thermal ramps, tightening float precision on continuous freeze clocks, and auditing every slider label for WCAG compliance required exhaustive fuzzing.
What's next for ColdChainGuard
Review the educational scenarios with rural immunization nurses and cold-chain logisticians, add multi-chamber facility layouts, and compare simulated holdover curves against published open-access passive-carrier test data.
Built With
- css3
- healthcare
- html5
- javascript
- node.js
- service-worker
- svg
- web-crypto-api