Building India’s first Tuberculosis Q&A Bot

Speech data

September 2023

Building India’s first Tuberculosis Q&A Bot

placeholder

The Challenge

Across India, there are about 2.4 million notified TB patients. In Karnataka the situation is particularly dire due to lack of effective communication and knowledge regarding tuberculosis diagnosis and mitigation - as of 2023, Karnataka’s TB prevalence to notification ratio is far higher than national average. This highlights a key gap in knowledge.

The Solution

Karya is working with the Karnataka Health Promotion Trust (KHPT) and USAID to create a voice-based tuberculosis knowledge app. Our goal through this project is to build a voice-assistant app for KHPT to target their core demographic - TB patients and caregivers. This smartphone application’s aim is to support the spread of knowledge and best practices for TB patients and their caregivers. It will allow patients to ask general knowledge questions about Tuberculosis in Kannada and get an intelligent response back.

For example:

A patient can ask “What types of food are okay to eat with my TB medicines?”.

And get the response: “You don’t have to worry too much about what you eat with your medication. Most things are okay, but it’s good to eat lots of Vitamins A, C and E. You can get this by adding more fruits and vegetables to your diet! Here are some examples of what you can add: orange, mango, sweet pumpkin and carrots, guava, amla, tomato, nuts and seeds.”

For a voice assistant application to be usable by a wide array of users across Karnataka, it must be sensitive and responsive to the various accents, dialects and ways of speaking. To build this app, Karya will be employing 100 individuals (50 men and 50 women) each in 10 districts across Karnataka to collect speech data using the Karya app.

With India's growing economy, the value of Kannada data has significantly increased. As a part of the Karya Promise and our goal of wealth distribution, we want to extend the value of these lucrative datasets to the workers. This means that once collected, validated, and used for the KHPT app, the same dataset will be available for sale on the Karya marketplace, with workers receiving royalties on any future data re-sales. This means that the data we are collecting has two avenues of impact. First, the data is used to develop a rigorous voice-based app in Kannada designed specifically for spreading TB knowledge. Second, the worker community has the potential to profit from data royalties on resales through the Karya’s Marketplace.