Sign-Based AI System for Inclusive Healthcare

Our Core Message

Welcome to the Sign-Based AI System for Inclusive Infectious Disease Screening and Diagnostics. We are developing a cutting-edge, dual-model Artificial Intelligence tool—fusing Large Language Models (LLMs) and Large Vision Models (LVMs)—to translate Ugandan Sign Language (USL) in real-time for critical infectious disease screening, including malaria, tuberculosis, and HIV/AIDS.

But this project is not just about writing code; it is about rewriting power dynamics. Rooted in the disability rights mandate of "Nothing About Us Without Us," our fundamental philosophy is that technology must be built with the community, not just for them. By treating lived experience as non-negotiable domain expertise, we ensure that the Deaf community serves as the active architects of their own inclusive futures. We are moving away from extractive tech models that treat disabled bodies as defective, and instead building an AI that adapts to natural language.

3

Critical Barriers Addressed

10+

Co-Creators & Team Members

4

Phases of Co-Creation

Our Founding Philosophy

Nothing About Us Without Us

Technology must be built with the community, not just for them. By treating lived experience as non-negotiable domain expertise, we ensure that the Deaf community serves as the active architects of their own inclusive futures.

Learn More About Our Philosophy

Breaking the "Speech-Locked" Clinic

Currently, outpatient departments in Uganda operate almost exclusively on spoken language and visual literacy, assuming a "Normative Patient" who can hear a shouted name in a crowded room, speak rapidly, and read complex English forms. This ableist design creates a "speech-locked" clinic that actively manufactures medical errors and excludes those who communicate differently.

Because of this systemic failure, Deaf patients face severe, life-threatening barriers: forced proxy reliance, diagnostic overshadowing, and the textual wall of performative consent.

Co-Creation Futures Framework

Instead of the standard, extractive AI pipeline of "take data, build model, deploy," our methodology is to "negotiate values, co-define evidence, test futures, and distribute power". By intentionally stress-testing our AI through speculative social simulations before it is built, we are engineering an assistive tool that respects privacy-by-design, prevents diagnostic errors, and restores the epistemic authority and dignity of the marginalized patient.

Explore Our Methodology

Join Us in Reclaiming Healthcare Futures

We are building technology that adapts to natural language, not the other way around.

Explore Our Methodology Meet the Co-Creators See the AI in Action

In partnership with: