MaaCheck

800,000 children die of pneumonia yearly. The tools to detect it exist — built for clinics, not mothers. MaaCheck puts WHO-validated screening in every mother's hands. Offline. Free. 3 minutes.

Comment

Inspiration

I grew up hearing stories about my grandmother delivering babies in a village where the nearest doctor was two days away by foot. She didn't have tools. She had instinct, experience, and a prayer.

That world hasn't disappeared. It has just moved from two days to two weeks.

Today, a community health worker visits a mother and her child every 14 days. She brings MUAC tapes, a respiratory timer, a danger sign checklist — the same tools that can detect pneumonia in under three minutes. She checks everything, marks her register, and leaves.

Then the fever starts on Day 3.

The mother watches her daughter's chest rise faster than normal. She watches her stop eating. She doesn't know if this is the kind of fever that passes — or the kind that kills. She has no tool. She has no number to call. She has nothing built for her.

800,000 children die of pneumonia every year. Not because the tools don't exist. Because every single tool — MUAC tape, respiratory timer, IMCI checklist — was built for the worker who visits once a fortnight. Not the mother who is there every minute.

That gap is what SDG 3.2 — ending preventable child deaths under five — is actually asking us to close.

We built MaaCheck so no mother has to guess.

What it does

MaaCheck is the first WHO IMCI-validated child health screening tool built for mothers — not health workers.

In three minutes, on any phone, fully offline, a mother gets one clear answer: "See a health worker TODAY" or "Your child appears healthy — monitor at home."

Three checks power that answer:

⚡ Danger Signs — 6 WHO IMCI general danger signs, one at a time, read aloud by voice. No reading required. No training required. Just two large buttons: Yes or No.

🫁 BreathCount™ — Our core innovation. The mother watches her child's chest and taps the screen once per rise for 60 seconds. The app counts, computes the rate, and compares it against WHO age-specific fast-breathing thresholds:

$$\text{Fast breathing} = \begin{cases} \geq 60/\text{min} & \text{under 2 months} \ \geq 50/\text{min} & \text{2–11 months} \ \geq 40/\text{min} & \text{1–5 years} \end{cases}$$

This is clinically identical to the method CHWs are trained to use with a stopwatch — it simply removes the need for training.

📏 BandCheck — The mother wraps the MUAC tape her ASHA worker already provides during home visits around her child's arm and selects the colour she sees. Red, yellow, or green. The app classifies against WHO SAM/MAM thresholds and recommends action.

Results are shareable via WhatsApp in one tap. All history saved locally. Nothing requires internet. Nothing requires an account.

How we built it

Every technical decision was driven by one question: will this work for a mother in a dimly lit room in rural Bihar with a squirming two-year-old?

React + Vite — Deploys as a static site with no backend. A CHW can share the link via WhatsApp and it opens instantly in a browser, no app store required.

Tap-based BreathCount, not camera detection — We built camera-based chest detection first. It looked impressive. Then we tested it in dim indoor light with a moving child and found real-world accuracy dropped below 70%. We scrapped it. The tap method — the same method CHWs are trained to use — works in any lighting, on any phone, every time.

Web Speech API — Reads every question aloud in English and Hindi, removing the literacy barrier entirely without requiring any third-party service or internet connection.

WHO IMCI 2024 thresholds hardcoded — The app is clinically accurate with zero connectivity. There is no server call between a mother's tap and the result.

The MUAC problem solved by the existing system — We didn't need to give mothers a new tool. CHWs already carry MUAC tapes on every visit. We made that tape meaningful between visits.

Challenges we ran into

Making clinical precision feel like a conversation. WHO IMCI is written for trained providers. Converting it into questions a first-time smartphone user answers correctly under stress — without making it clinically incomplete — was the hardest design problem we faced.

A hidden CSS collision that broke the core demo. The landing page and the tap counter shared the same CSS animation name (breathe). The later definition silently removed the centering transform from the tap button — the button was visually displaced and untappable. The fix was one line. Finding it took two hours.

Building a risk engine that never reassures falsely. A child with normal nutrition but fast breathing is not healthy. Our combined risk engine always surfaces the worst result — it never averages across checks and it never lets a green result bury a red one.

Accomplishments that we're proud of

Every clinical threshold is WHO IMCI 2024 exact. Not approximate. Not inspired by. Identical.

A mother who cannot read can complete a full screening. Voice reads every question. Colour zones replace measurements. Symbols replace text. The entire experience was designed for the user who has never seen a medical form.

The CHW collaboration model is deployable today. No behaviour change required from health workers. No new infrastructure. The app slots into the existing system — it extends it, not replaces it.

Full bilingual support — English and Hindi — from the ground up. Not translated after the fact. Both languages feel native because both were written as first languages.

What we learned

We learned that the gap between existing tools and the people who need them most isn't technical — it's assumed.

Every developer who built a health screening tool assumed a trained operator. That assumption was so fundamental, no one even identified it as a choice. When we questioned it, a completely unaddressed design space opened up.

We also learned that simplicity is a clinical feature. A confused mother who answers a question incorrectly produces a wrong result. A wrong result is worse than no result. Every time we made something simpler, we made it more accurate for the actual user — not less.

What's next for MaaCheck

Phase 1 — Language expansion (Months 1–2) Swahili, Hausa, and Amharic — the three languages covering the majority of Sub-Saharan Africa's under-five mortality burden, and the region where GNEC's 1,600 subsidiaries operate most actively.

Phase 2 — CHW Dashboard (Months 3–4) Aggregate screening data across a health worker's assigned families. Which children triggered high-risk results this week. Which households haven't screened in 10 days. A village-level early warning system built on top of every mother's individual screenings.

Phase 3 — National Programme Integration (Months 5–12) Formal integration with India's ASHA programme, Kenya's Community Health Worker network, and Ethiopia's Health Extension Programme — the field systems that already reach the mothers MaaCheck is built for.

Phase 4 — Beyond Pneumonia The same architecture — offline, voice-guided, mother-facing, WHO-validated — applies directly to SDG 3.1 (maternal mortality), SDG 3.3 (communicable disease screening), and immunisation tracking. MaaCheck is a platform, not a single tool.

My grandmother delivered babies with instinct and a prayer. Amara's mother deserves better than that. So does every mother like her — and there are hundreds of millions of them.

Every mother deserves to know.

Built With

Share this project:

Updates