Satipo, Amazonian Peru 🌿

Predictive AI That Acts
Before the Mosquito Does

The first IoT + AI system built by Amazon students to predict dengue larval risk and automatically deploy Bti β€” a WHO-approved biological larvicide β€” before mosquitoes emerge.

🟑 MEDIUM RISK β€” System on Alert
Last Bti deployment: 2 hours agoTotal this session: 8
273,847
dengue cases in Peru β€” 2024
S/ 107M
wasted on fumigation that fails
0 systems
like DengueStop existed before us
Live Sensor Dashboard

Real-time readings from the Amazon

Updated every 3 seconds

Temperature
26.0
Β°C
Thresholds: 25 / 28 Β°C
Humidity
78.3
%
Thresholds: 70 / 80 %
🟑 MEDIUM RISK β€” System on Alert
Last Bti deployment: 2 hours agoTotal this session: 8
7-Day Climate History β€” Satipo

Temperature & Humidity Trend

Model trained on SENAMHI public climate data β€” Satipo, JunΓ­n, Peru (2020–2024)
Community Field Survey

Boots-on-the-ground inspection β€” Satipo

Students physically inspect water containers weekly. AI recommends, humans confirm.

Containers inspected
47
With larvae
12
Larval Index (IR%)
25.5%
⚠ HIGH (WHO threshold: >5%)
Bti deployments this week
8

IR% = (containers with larvae / total inspected) Γ— 100

Jr. Amazonas
πŸͺ£ Bucket
HIGH
Larvae
23
Bti deployed βœ“
Near central market
πŸ”΄ Tire
HIGH
Larvae
11
Bti deployed βœ“
School yard
πŸͺ΄ Flowerpot
LOW
Larvae
0
Clean βœ“
Jr. JunΓ­n
πŸ’§ Water tank
MEDIUM
Larvae
8
Bti deployed βœ“
Residential block
🧺 Tray
LOW
Larvae
0
Clean βœ“
Jr. PerΓΊ
πŸ›’οΈ Barrel
HIGH
Larvae
17
Bti deployed βœ“
Church area
πŸͺ£ Bucket
MEDIUM
Larvae
3
Bti deployed βœ“
Plaza central
πŸͺ΄ Flowerpot
LOW
Larvae
0
Clean βœ“
Jr. Ucayali
πŸ›’οΈ Barrel
MEDIUM
Larvae
5
Bti deployed βœ“
ZoneContainerLarvaeRiskBti Deployed
Jr. AmazonasπŸͺ£ Bucket23HIGHYes βœ“
Near central marketπŸ”΄ Tire11HIGHYes βœ“
School yardπŸͺ΄ Flowerpot0LOWβ€”
Jr. JunΓ­nπŸ’§ Water tank8MEDIUMYes βœ“
Residential block🧺 Tray0LOWβ€”
Jr. PerΓΊπŸ›’οΈ Barrel17HIGHYes βœ“
Church areaπŸͺ£ Bucket3MEDIUMYes βœ“
Plaza centralπŸͺ΄ Flowerpot0LOWβ€”
Jr. UcayaliπŸ›’οΈ Barrel5MEDIUMYes βœ“
How DengueStop Works

From sensor to action in 5 steps

1
🌑️
SENSE
DHT22 sensor reads temperature + humidity every 60s
2
🧠
PREDICT
AI model compares live readings to 4 years of SENAMHI historical climate patterns from Satipo
3
⚠️
DECIDE
If TΒ° β‰₯ 25Β°C AND HR β‰₯ 70% sustained for 30 minutes: HIGH RISK confirmed β†’ trigger deployment
4
πŸ’§
ACT
ESP32 activates micro-pump β†’ releases 1mg/L Bti into water container automatically
5
🦟
ELIMINATE
Bti destroys larvae by ingestion in 24–48h. ⚠ NOTE: Bti acts ONLY on larvae β€” NOT on eggs or pupae. We disclose this limitation openly.
Responsible AI

We disclose what our system can β€” and can't β€” do

Risk Identified

False positives may trigger unnecessary Bti release. False negatives may miss real outbreaks due to limited local training data.

How We Manage It

Students physically inspect containers weekly using a standardized field protocol (IR% calculation). AI recommends β€” humans confirm before any large-scale deployment.

Human in the Loop

Community health workers review the dashboard weekly. Families decide whether to install the device in their home. The AI never acts alone.