Ward-level dengue forecasts that tell city spray crews where to go next
TL;DR Rainlag gives a city's health department a weekly forecast of which wards are likely to see dengue rise over the next four to eight weeks, built from rainfall, temperature and the city's own case reports. Municipal corporations pay a yearly fee. It works because spray crews are limited and timing is everything.
India recorded 289,235 dengue cases and 485 deaths in 2023, and 233,519 cases in 2024[1]. Those are reported cases only, and an ICMR-led study in Goa notes that dengue is substantially under-reported and that passive data understate the burden[2].
City vector-control teams mostly react after cases climb. The same Goa study linked heavier rainfall to more transmission two to three months later[2], so the warning signal is there well before the hospital rush. Most cities just don't use it ward by ward.
What it does, in regulatory terms: public-health decision support for vector control. Rainlag gives no diagnosis, handles no patient records and makes no treatment claims. It works on case counts already aggregated by ward.
The seed pitched one platform predicting any outbreak from many data streams. That needs data no startup can get and overlaps the government's own IHIP surveillance system[7]. So I narrowed it to one disease with a strong weather signal and a clear buyer.
Each week Rainlag pulls weather data, the city's ward-wise dengue counts and, where a city signs data-sharing letters with private labs, their aggregated positivity. It ranks wards by forecast risk, shows the uncertainty honestly and hands the vector-control officer a short list of these ten wards for these weeks. After each season it publishes how the forecasts did against actual cases.
The users are the city's vector-borne disease officer and the ward health inspectors. The payer is the municipal corporation's health department, through an annual contract. State vector-borne disease programmes are a second buyer for district-level versions. There's no public price for this kind of service in India, so price it as a small share of what a city spends on fogging and spraying and confirm that number in interviews.
Indian research teams have shown the method works locally. IITM Pune built a model with a two-month lead time for Pune in 2025[3], and Kerala's Institute of Advanced Virology launched a district-level warning system in October 2026[4]. Both say local health data is the bottleneck, which is exactly the gap a vendor can fill city by city.
Nobody publishes what Indian cities spend on dengue forecasting. The case for paying is the cost of spraying in the wrong place. A 2026 model for Ahmedabad found sustained residual spraying could cut modelled incidence by over 80%[5], so aiming that spraying better matters a lot.
On the adjacent side, One Canadian firm sells outbreak intelligence to governments worldwide[6].
IITM Pune[3] and Kerala's IAV[4] are research or state programmes, not services a city can buy. IHIP is the national reporting platform and not a forecasting tool[7]. That firm tracks over 190 diseases for governments and companies[6], but it works at country and city level rather than ward-by-ward spray plans.
The gap is a vendor that runs the forecast every week for one city's crews.
| Difficulty | Moderate (46/100) |
| Starting capital | ₹60 L–1.2 Cr |
| Time to launch | 9–15 months |
The modelling is well studied. The hard parts are getting clean ward-level case data out of a municipal corporation, keeping it flowing every week and selling to government buyers who move slowly and buy through tenders. Expect the first season to be a free trial, whether you like it or not.
No health licence is needed while Rainlag stays at ward-level decision support and never touches patient records.
Building this outside India? The categories above usually have a local equivalent, so check with your own regulator.
Don't let the model replace field judgement. It suggests where to look and gives no diagnosis, so officers should weigh it against what their inspectors see on the ground.
Want to work out the AI layer for Rainlag? Talk to Yash →
Here's what can go south -
I like this for a founder with epidemiology or public-health data experience and the patience for government sales. If you want quick revenue then this isn't it.
Prove one thing first - using one city's past five seasons, show the model would have flagged the worst wards at least four weeks early.
My estimate is ₹60 L–1.2 Cr to launch, covering data engineering, epidemiology staff and a long first season with one city.
No health licence while you use only ward-level counts and give no diagnosis. Expect tender rules, and act as the city's data processor under the DPDP Act.
Yes, wherever dengue tracks rainfall. Check local health-data rules and how cities buy.
The free Rainlag validation kit has the line-by-line capital breakdown, the effort scoring, the data-protection and procurement checklist, a 14-day validation sprint with pass/fail targets, a 90-day launch plan, 10 interview questions for city health officers, and 5 assumptions to test. Download the kit →
Thinking about building Rainlag or something close to it? The make-or-break work is getting one city to share weekly ward data and stick with you through a full season, and I'm happy to help you plan that approach. If you need a hand, you know where to find me. Talk to Yash →
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.