Makerere University • Clinical AI

Democratizing Oral Healthcare Through Video-Powered Vision Models

Engineered at Makerere University. UGDent is an enterprise-grade, mobile-first diagnostic infrastructure that abandons fragile static-image analysis in favor of real-time, video-embedded Large Vision Models to detect and localize dental diseases in low-resource clinical settings.

Video-Powered AI
Sub-2 Second Inference
Explainable AI
Uganda-Built
10,500+ Clinical Images Across 6 dental pathology categories
<2s Inference Time Real-time video stream analysis
87% Accuracy CLIP Vision Transformer validation
About the Platform

What is UGDent?

UGDent is a clinical decision-support API and mobile application designed for real-time intraoral screening. Unlike conventional AI models that require perfectly lit, high-resolution static photos, UGDent processes continuous video streams captured directly from standard smartphones. By utilizing a highly optimized, fine-tuned CLIP Vision Transformer (ViT) architecture deployed over a persistent WebSocket connection, the platform delivers sub-2-second diagnostic inference for conditions such as caries, gingivitis, calculus, and oral ulcers, regardless of camera shake or variable lighting.

Community Health Workers

Empowering frontline workers to perform accurate oral health screenings

Tele-Dentistry Platforms

Scalable, low-latency API for automated dental diagnostics

Dental Clinics

Real-time, explainable "second opinion" for routine examinations

Real-Time Video Processing

Continuous intraoral video stream analysis for robust diagnostics

CLIP Vision Transformer

Fine-tuned ViT architecture for semantic understanding

Sub-2 Second Inference

WebSocket-powered ultra-low-latency diagnostic delivery

Explainable AI

Attention maps highlighting exact disease localization

The Problem We Solve

Workforce Shortage

Uganda faces a crippling oral health workforce shortage, with approximately 1 dentist per 150,000 people, leaving rural populations without access to diagnostic expertise.

Static Image Failure

Imported AI solutions rely on perfectly lit, high-resolution static photos. In rural Ugandan clinics, they fail dramatically due to motion blur, saliva glare, and variable lighting.

Preventable Escalation

Preventable conditions like tooth decay and periodontal disease escalate into severe disability and economic hardship without early diagnostic expertise.

Why Uganda, Why Now

Our Strategic Market Opportunity

Uganda is uniquely positioned to leapfrog traditional diagnostic bottlenecks by leveraging the rapid penetration of mobile technology. As the burden of non-communicable oral diseases grows, we cannot wait decades to train enough dental surgeons to cover every district.

UGDent operationalizes the "Buy Uganda, Build Uganda" philosophy within health-tech. By engineering a frugal, video-capable Large Vision Model locally, we ensure that Ugandan health systems can deploy clinical-grade, context-aware screening tools without paying exorbitant licensing fees for opaque, imported medical software that fails in our operating environments.

Locally Engineered Mobile-First Clinically Validated Data Sovereign

Impact at a Glance

1:150,000 Dentist-to-patient ratio
10,500+ Clinical images validated
<2s Diagnostic inference time
87% CLIP ViT accuracy

Founders & Strategic Leads

Kakuru Conrad Akankwasa

Co-Founder & Technical Lead - ML

Specializing in Large Vision Models, CLIP architectures, and real-time video inference. Leads the engineering of the UGDent diagnostic engine.

Tulinawe Ssebuhinja

Co-Founder & Lead Engineer - Deployment

Specializing in WebSocket architecture, FastAPI deployment, and React Native mobile development. Oversees clinical integration and MLOps.

Dr. Ggaliwango Marvin

Strategic Advisor & AI Mentor

Providing oversight on model optimization, research-to-product commercialization, and responsible AI governance in healthcare.