How bhavya173045/med_triage_pipeline works
MED_TRIAGE_PIPELINE is a Streamlit hospital triage portal with a separate clinical-trial data and model-training pipeline. Staff enter clinical notes and a priority; the app loads the trained model, runs NLP analysis and risk fusion, checks for priority mismatches, and renders decision-support results. Patient history is read from a CSV. Separately, a pipeline fetches trial studies, preprocesses them, and trains the model artifact used by the portal. The README documents dashboard visualization and stress testing; source evidence shows the test hook, but not a separate patient-history persistence workflow.
Main entry points: app.py Triage App · run.py Pipeline Runner · scraper.py Trial Scraper · processor.py Trial Preprocessor
Components 13
Triage Portal
- Triage App
src/app.py - Result Views
src/streamlit_components.py
Risk Assessment
- Risk Engine
src/risk_engine.py
Model Pipeline
- Pipeline Runner
run.py - Trial Scraper
src/scraper.py - Trial Preprocessor
src/processor.py - Model Trainer
src/classifier.py
Clinical Data
- Patient History
data/patient_records.csv - Trained Model
models/clinical_trials_model.pkl - Processed Trials
data/processed/cleaned_trials.csv
Safety Checks
- Robustness Test
src/tests_robust.py
Other components
- Hospital Staff
- ClinicalTrials.gov
Connections 13
- Hospital Staff → Triage App: enters triage
README.md - Triage App → Patient History: reads records
src/app.py - Triage App → Trained Model: loads model
src/app.py - Triage App → Risk Engine: analyzes and scores
src/app.py - Triage App → Result Views: renders results
src/app.py - Triage App → Robustness Test: runs checks
src/app.py - Robustness Test → Risk Engine: tests scoring
src/tests_robust.py - Pipeline Runner → Trial Scraper: fetches studies
run.py - Trial Scraper → ClinicalTrials.gov: requests studies
src/scraper.py - Pipeline Runner → Trial Preprocessor: transforms data
run.py - Trial Preprocessor → Processed Trials: writes rows
src/processor.py - Pipeline Runner → Model Trainer: trains model
run.py - Model Trainer → Trained Model: saves artifact
src/classifier.py