Evolution of the Oral Cancer Screening App

India has one of the highest oral cancer burdens in the world — 77,000 new cases and 52,000 deaths annually. Most are diagnosed too late, when treatment options narrow and survival rates drop.

The reasons compound: rural areas lack healthcare infrastructure, awareness is low, and habits like tobacco use are widespread. But the central issue IISc identified was simple — there was no mechanism for early detection at scale.

To address this, IISc organised screening camps in rural and remote areas. The screeners would be ASHA health workers — frontline community workers with no clinical training. They needed a tool that could guide them through the screening process and use AI to flag potential cases. That tool became Aarogya Aarohan.

Aarogya Aarohan is part of AI-COE (Artificial Intelligence — Centre of Excellence), an Indian government initiative to apply AI for public good. The project was sponsored by IISc and Niti Aayog, governed by ICMR, and built by our team at Triveous.

I joined after the first phase had shipped and been pilot-tested at AIIMS Delhi with ASHA workers and nurses. The usability testing surfaced real problems — and solving them became my responsibility across the next two design phases.

The Delhi pilot revealed who we were really designing for — and the constraints were severe.

When a call came in mid-screening, the app crashed. The worker had to restart the entire process from scratch — re-entering patient details, retaking photos, re-running the AI inference. In a camp setting with dozens of patients waiting, that wasn't just frustrating — it meant people went unscreened.

First, a draft-save mechanism. If a screening was interrupted — by a phone call, a battery warning, anything — the app preserved the incomplete session automatically. Workers could pick up exactly where they left off. No re-entry, no lost photos, no repeated AI inference.

I explored multiple layout options, evaluated each for usability on small screens and feasibility with the engineering team, then moved from wireframes to high-fidelity screens with detailed technical guidelines. Explaining offline-first and draft persistence to engineers required multiple sessions — the behaviour was non-trivial to implement correctly.

The first redesign focused on trust and reliability. Drafts and synced cases lived in separate tabs. Navigation stayed flat — no nested screens, no buried actions. Everything a worker needed was one tap away.

The Delhi pilot was a starting point, not a conclusion. To understand how the app performed in the conditions it was actually built for, the IISc team organised two screening camps in rural Krishnagiri, Tamil Nadu — one in Mallapadi and one in Kanthikuppam, with support from the state government.

The insights from Krishnagiri changed the direction of the project. The existing layout couldn't support the features workers actually needed — follow-up tracking, a dashboard for camp progress, notifications, and profile management. Patching the current design wasn't an option.

I rebuilt the information architecture and sitemap from scratch to accommodate the new features — Dashboard, Cases, Tasks, Profile, and Notifications — while keeping the core screening flow intact. Daily huddles with the IISc team in Bangalore, who were running parallel usability research, kept the design aligned with incoming findings.

For the design system, we moved away from ad-hoc components and built a purpose-built system using Material Design as the foundation — optimised for the low-end Android devices ASHA workers actually used. For the backend admin panels, we continued with Carbon Design System, which we were already using for MIDAS.

I worked through multiple rounds of sketches and lo-fi wireframes before moving to high fidelity. The final screens reflected everything we'd learned — from the Delhi pilot, the Krishnagiri camps, and ongoing IISc research.

Three core surfaces — a dashboard for camp-level progress and follow-up tasks, a cases screen separating drafts from submitted and synced screenings, and a tasks section for tracking new, pending, and completed follow-ups.

Aarogya Aarohan replaced a fully manual, paper-based screening process. ASHA workers now complete screenings faster with AI-inferred results generated instantly. The draft-save mechanism eliminated data loss from phone interruptions — a problem that previously forced workers to restart entire screenings.

The app is deployed across 5 states — Karnataka, Tamil Nadu, Delhi, Assam, and Uttar Pradesh — with hundreds of active ASHA workers using it daily.