Challenge
Biliary atresia and related conditions can first appear as changes in stool colour. The product needed to convert parent-captured photos into a reliable screening signal, without requiring any medical knowledge from users.
Solution
We built a React Native app that guides parents through image capture and quality checks, then a Python-based computer-vision model, trained on clinically annotated images, analyses each submission against an MDR-aware validation workflow.
Results
- Supports earlier referral for high-risk infants
- Delivers scalable screening with less regulatory rework
- Removes reliance on parents' clinical judgement
- Provides a validated, production-ready capture pipeline
Technology
React Native · Python · TensorFlow · Computer Vision
Powered by DAC Digital