Walelign/amharic-srh-chatbot · Apply for community grant: Academic project (gpu and storage)

GenAI-based Multimodal and Multilingual Sexual and Reproductive Health and Rights (SRHR) Chatbot in Low-resource Languages

Project Summary

This project proposes the development and deployment of a Generative AI-powered chatbot designed to improve access to Sexual and Reproductive Health and Rights (SRHR) information in low-resource, underrepresented languages. The chatbot will integrate multimodal (text, voice, image) and multilingual capabilities, ensuring inclusive and culturally relevant engagement for underserved populations, particularly youth, women, and rural communities.

Problem Statement

In many regions, especially across Sub-Saharan Africa and other low-resource settings, access to accurate and respectful SRHR information is hindered by language barriers, digital illiteracy, social stigma, and infrastructural gaps. Traditional health communication channels often exclude non-dominant language speakers and fail to address culturally specific SRHR concerns. The result is poor awareness, unsafe practices, and preventable health risks.

Proposed Solution

We will design and implement a GenAI-based SRHR chatbot capable of:

Interacting in low-resource African languages Supporting both text and voice interfaces. Addressing SRHR topics in a non-judgmental, youth-friendly tone Integrating early warning features for common or dangerous symptom patterns. Providing users with localized health resources and connecting them to nearby services. Aggregating anonymized insights for data-driven policy and service improvement

Social Impact Focus

Improving Access to SRHR Information

    Breaking language and digital literacy barriers with voice and vernacular support
    Empowering young people, especially girls and women, with knowledge to make informed decisions

Reducing Stigma

    Enabling private, judgment-free conversations about sensitive SRHR issues
    Promoting culturally contextual education on taboo subjects

Early Warning System

    Detecting trends in symptom reporting (e.g., recurring UTIs, menstruation irregularities, STIs)
    Alerting health workers and communities to potential health crises

Data-Driven Health and Policy Decisions

    Offering anonymized data insights to support public health strategies and interventions
    Identifying gaps in service delivery and resource allocation

Target Audience

Youth and adolescents (esp. girls and women)
Speakers of underrepresented languages
Rural and underserved populations
Community health workers and local clinics

Sustainability & Scalability

The system will be developed using open-source LLM platforms with community training datasets, allowing local adaptation and continuous improvement. The model and tools can be extended to other health domains (e.g., maternal health, mental health), and deployed across various countries with similar language and infrastructure profiles.

Please find the demo below:
https://huggingface.co/spaces/Walelign/amharic-srh-chatbot