huraimaali/malnutrition-screening-children

Child Malnutrition Risk Screening

Real-time multi-task deep learning system predicting stunting, wasting, and underweight risk in children under 5, using household and demographic factors — with Monte Carlo Dropout uncertainty estimation and SHAP explainability.

Trained on DHS (Demographic and Health Survey) Bangladesh 2022 data, using WHO Child Growth Standards for ground-truth z-scores.

This is a research/portfolio project and screening aid — it does not replace clinical assessment or physical measurement.

See the full project writeup and methodology in the main repository README.