About

Public health data, read at the level of the household and the district.

Focus

Machine learning for maternal & child health outcomes (DHS data)
Spatial epidemiology — hotspot detection, kriging, residual mapping
Explainable AI (SHAP) for clinically meaningful predictors
Accessibility tools for Ethiopian languages

Tools

Python Stata 17 ArcGIS 10.8 XGBoost SHAP scikit-learn Google Colab Survey-weighted regression GEE R

Background

MSc Health Data Science

In progress · Debre Markos University

Applied machine learning, spatial analysis, and biomedical image processing for Ethiopian public health problems, supervised by instructors including Mulat Belay.

MPH coursework

Epidemiology & Health Data Science tracks

Longitudinal data analysis (LMM, GLMM, GEE), epidemiological study design, and qualitative research methods.

Childhood stunting research

EDHS 2016 & 2024–25

Spatial analysis of childhood stunting using ArcGIS (Moran's I, Getis-Ord Gi*, kriging) and an ML pipeline in progress for a publishable follow-up paper.

Institutional delivery prediction

EDHS 2024–25 Births Recode

XGBoost vs. logistic regression (test AUC 0.88) with SHAP explainability and spatial hotspot analysis from out-of-fold residuals.

Amharic speech-to-text

Accessibility project

Real-time transcription tool for hearing-impaired Amharic speakers, built on faster-whisper.