JolSetu — the AI That Reads a 20-Taka Arsenic Test Kit (So Families Can Find Safe Water)
Description and use
On 943 real kit-vs-lab comparisons, human reading caught 99.3% of unsafe wells but wrongly condemned 37% of safe ones. JolSetu's standardized AI reading attacks exactly that error. The prototype uses 6,593 real wells from the Araihazar upazila survey via CUAHSI HydroShare.
Problem
20 million Bangladeshis drink water above the national arsenic standard (50 µg/L); 45 million above the WHO guideline. Testing is once-in-decades, readings are human color-chart guesses, and when a well fails, families have no way to find the nearest verified safe one.
Outcome
On 943 real kit-vs-lab comparisons, human reading caught 99.3% of unsafe wells but wrongly condemned 37% of safe ones. JolSetu's standardized AI reading attacks exactly that error. The prototype uses 6,593 real wells from the Araihazar upazila survey via CUAHSI HydroShare.
Workflow
- Photograph the Gutzeit reaction pad next to the white reference strip using a mobile phone.
- On-device color science reads the chart using white-reference normalization and CIELAB matching.
- The AI speaks the arsenic verdict in Bangla.
- Check-in is HMAC-signed, GPS-geofenced at 30 m pass/flag/reject at 2× radius, duplicate-checked per day, and reviewable.
- A logistic risk model flags where model and map disagree; the crowd tests there next.
- Distance, compass direction, and walking time to the nearest verified safe wells are computed for routing.
Evidence
20 million Bangladeshis drink water above the national arsenic standard (50 µg/L); 45 million above the WHO guideline — Flanagan et al., Bulletin of the WHO, 2012
24,000+ adult deaths every year are attributed to arsenic above 50 µg/L, plus up to 19,000 more in the 10–50 µg/L band — same source
WHO called it 'the largest poisoning of a population in history' (Smith et al., 2000)
Bangladesh ran massive national campaigns — 10M+ tubewells tested (BAMWSP 2000–2005, ARRMP 2021–23). But testing is once-in-decades, readings are human color-chart guesses, and when a well fails, families have no way to find the nearest verified safe one.
on 943 real kit-vs-lab comparisons in the repo's dataset, human reading caught 99.3% of unsafe wells but wrongly condemned 37% of safe ones. Standardized AI reading attacks exactly that error.
Built as a university-track prototype for the Xylem Global Student Innovation Challenge 2026, water-quality problem statement.
All 6,593 wells in the prototype are real measurements from the Araihazar upazila survey (Narayananj, Bangladesh), via CUAHSI HydroShare: DOI 10.4211/hs.8e1373d87419447c945625af13f0a2ea. The repo ships an EVIDENCE.md with a claim→source map, every DOI verified live, plus a model card.
The colorimetry assumes the standard kit's chemistry and reasonable lighting; the risk model is deliberately simple and its AUC is published; the map's trust model depends on review participation. This is a prototype with a real evidence base — not a deployed system.
- Record date: 2026-10-21T07:56:19.955Z