Robotics1025/uganda-integrated-care-intelligence-system

`UGICIS is a missingness-aware multi-task clinical AI system designed for Ugandan healthcare environments. The project predicts hypertension risk, treatment urgency, and intelligent clinical referral pathways for HIV patients using real Ugandan clinical data, explainable AI, and offline-capable deployment strategies.` # Uganda Integrated Care Intelligence System (UGICIS) ## A Missingness-Aware Multi-Task Clinical Intelligence System for HIV-Hypertension Risk Prediction and Clinical Pathway Support in Ugandan Healthcare Environments --- # Author Information **Principal Investigator/Evaluator/Dataset Relevance/ML/DL Structure:** Mugole Joel/Shadia/Elizabeth/Keith/Allan **Academic Programme:** Bachelor of Science in Computer Science (Year 2) **Institution:** Makerere University **Location:** Kampala, Uganda **Research Domain:** Artificial Intelligence for Healthcare / Digital Health / Clinical Machine Learning **Project Type:** Research + Deployable Clinical AI System **Target Environment:** Ugandan Public Healthcare Facilities. --- # Abstract Uganda is currently facing a dangerous dual disease burden in which communicable diseases such as HIV/AIDS coexist with rapidly increasing non-communicable diseases, particularly hypertension and cardiovascular complications. Patients receiving antiretroviral therapy (ART) are now surviving longer and increasingly developing hypertension, stroke risk, and metabolic disorders. However, healthcare facilities in Uganda operate under severe staffing constraints, inconsistent documentation practices, and limited digital infrastructure. This project proposes the Uganda Integrated Care Intelligence System (UGICIS), a missingness-aware multi-task clinical machine learning system designed specifically for low-resource African healthcare environments. UGICIS uses real Ugandan HIV-hypertension clinical data to simultaneously predict hypertension presence, hypertension severity, urgent treatment need, and clinical referral pathways. The core scientific contribution of the project is the MissAware-MTL architecture — a Missingness-Aware Multi-Task Learning framework that treats incomplete clinical records not as noise to be discarded, but as informative clinical signals reflecting real-world healthcare constraints. The system integrates: * FT-Tra …