New AI Diagnostic Tool Aletheia Offers Offline Support for African Healthcare

A new clinical decision support system, Aletheia, has been developed to address the severe shortage of specialist medical expertise in sub-Saharan Africa. With physician-to-patient ratios critically low, particularly in rural areas, existing AI diagnostic tools are often unusable due to their reliance on stable internet and high-end hardware.

Aletheia tackles these challenges by being an offline-first system, making it suitable for low-resource healthcare settings. It is built on the Qwen2.5-3B-Instruct model, fine-tuned with QLoRA on a dataset of 27,000 clinical reasoning samples. This dataset specifically focuses on 50 diseases common in East Africa, enhancing its relevance to the region.

Evaluations show promising results, with an 80.0% Top-1 diagnostic accuracy and 100.0% Top-3 accuracy across various clinical case categories. Crucially, Aletheia meets the memory constraints of the Africa Deep Tech Challenge 2026, demonstrating its practicality for deployment on standard hardware without requiring cloud infrastructure. This innovation could significantly enhance primary care diagnostics in remote African regions.