cervinet-delta

# CERVINet-DELTA: A Multimodal Temporal Framework for Labor Induction Prediction > This repository provides the official implementation of **CERVINet-DELTA**, a probabilistic deep learning framework for individualized labor outcome prediction and decision support. It leverages multimodal data (structured EHR, uterine signals, ultrasound imagery) and integrates temporal modeling, attention mechanisms, and utility-driven policy optimization. ## 📌 Highlights - 🔬 **CERVINet**: Stochastic latent dynamics model for cervical progression- 🧠 **Multivariate Outcome Prediction**: Predicts delivery time, mode, and neonatal conditions- 🕒 **DELTA Layer**: Utility-based policy advisor with real-time risk stratification- 💡 **Explainable AI**: Attention visualizations and feature saliency- 🧪 Validated on large-scale retrospective **Dinoprostone** and **Multimodal Labor** datasets --- ## 📂 Project Structure ```bashCERVINet-DELTA/│├── data/ # Dataset loaders and preprocessing scripts│ └── ...│├── models/ # Core model modules (CERVINet, DELTA, etc.)│ ├── cervinet.py│ ├── delta_policy.py│ └── ...│├── utils/ # Helper functions (metrics, visualizations)│├── train.py # Training script├── evaluate.py # Evaluation script├── requirements.txt└── README.md