drankitjoshi/Post-NAHT-Prostate-Cancer-Outcome-Predictor

NAHT Outcome Predictor

Gradio deployment package for the rebuilt NAHT prediction suite.

Included

• —Deployable BCR model: bcr_clinical
• —Optional pathology-aware BCR model: bcr_postop
• —Margin, node, and exploratory pCR prediction
• —Clinical Cox model for BCR-free survival

Required project folders

Keep these folders beside app.py when deploying:

• —model_rebuild_outputs/
• —manuscript_package/
• —app_reports/ (will also be created automatically if missing)

Local run

python app.py

Notes

• —The app is intended for research and decision-support prototyping.
• —pCR remains exploratory because of low event count.
• —For production clinical deployment, add authentication, audit logging, and secure hosting.