All work
Maternal Health Risk Prediction
Graph neural networks plus gradient boosting to flag high-risk pregnancies — with explainable predictions.
Built bySadia- type
- ML research
- model
- GCN embeddings → XGBoost
- interpretability
- SHAP · t-SNE
- language
- Python
Overview
A machine learning research project that predicts high-risk pregnancy outcomes, including preeclampsia, by combining graph representation learning with classical and deep classifiers. Clinical data — blood pressure, blood sugar, BMI and biomarkers — is used to build patient similarity graphs; node embeddings learned with GCN, GraphSAGE and GNN variants feed downstream classifiers. The main pipeline pairs a GCN with reconstruction and domain-adaptation layers with a tuned XGBoost classifier, explained through SHAP and t-SNE.
What it delivers
- Systematic comparison of graph-based hybrid models for maternal health risk
- Interpretable pipeline — predictions come with SHAP explanations
Features
Patient Similarity Graphs
- Graphs built from combined maternal health and preeclampsia datasets
- Clinically similar patients connected as neighbours
- Node embeddings learned with GCN, GraphSAGE and GNN variants
Hybrid Models
- Embeddings compared across XGBoost, LightGBM, Random Forest, MLP and LSTM
- GCN with reconstruction and domain-adaptation layers
- Hyperparameter tuning and cross-validation throughout
Explainability
- SHAP feature importance for every prediction
- t-SNE views of the learned embeddings
- Extended experiments on synthetic data
Want something like this built?
Tell us what you have in mind and we will come back with a scope, a timeline and a number.