JosephWalusimbi-eng/Aletheia

Offline-First Clinical Decision Support AI for Sub-Saharan Africa # Aletheia — Offline-First Clinical Decision Support AI > *From the Greek ἀλήθεια — truth, disclosure. The revealing of what is hidden.* **Aletheia** is an offline-first clinical decision support system designed for frontline healthcare workers in district hospitals and health centres across sub-Saharan Africa. It runs entirely on a standard 8 GB laptop with no internet connection, providing ranked differential diagnoses, investigation recommendations, red flag identification, and clinical reasoning for 50 disease conditions prevalent across Africa. --- ## The Problem In Uganda, one doctor serves approximately 25,000 patients. A clinical officer at a district hospital may see 100 patients in a single day — fewer than 5 minutes per patient — to take a history, examine, diagnose, and treat. Existing AI diagnostic tools require cloud servers, fast internet, and expensive hardware. None of these are reliably available at the point of care in rural Africa. The clinicians who need AI support the most have the least access to it. ## The Solution Aletheia runs the entire clinical reasoning pipeline **on-device**: - ✅ **No internet required** — ever, at inference time - ✅ **No GPU required** — runs on CPU only - ✅ **1.80 GB model file** — fits on a USB drive - ✅ **~3,880 MB peak RAM** — well within the 8 GB ADTC budget - ✅ **Web UI + CLI** — browser interface or terminal - ✅ **50 clinical conditions** weighted for African disease epidemiology - ✅ **8 reasoning task types** — differential, tests, severity, management, red flags --- ## Hardware Target (ADTC 2026 Standard Laptop) | Spec | Value | |------|-------| | CPU | Intel Core i5 10th–12th gen | | RAM | 8 GB DDR4 | | Storage | 256 GB SSD | | GPU | None (integrated graphics only) | | OS | Ubuntu 22.04 LTS | | Internet | Not required | --- ## Installation ### Step 1 — Clone the repository ```bash git clone github.com cd Aletheia ``` ### Step 1b — Install Python …