Add final Guardian Monitor pipeline and backend alert bridge

[Tanvir-ctrl1]

Description

Added the final Guardian Monitor alert pipeline and backend bridge integration.

This update includes a cleaned final pipeline that classifies synthetic nurse notes into emotional/clinical tags, combines the text signal with vitals-based risk scoring, detects cross-modal anomalies, and produces Low/Medium/High alert outputs with nurse-readable explanations.

A separate backend bridge script was added to keep the main pipeline stable while allowing generated Medium/High alerts to be posted to the Guardian backend API.

Changes

• Added guardian_pipeline_final.py

  • Runs the final emotional tagging and alert pipeline
  • Generates baseline and LSA predictions
  • Produces evaluation report, prediction CSVs, and dashboard-style text cards

• Added guardian_backend_bridge.py

  • Runs the existing pipeline
  • Reads predictions_baseline.csv
  • Logs into the Guardian backend API
  • Sends Medium/High generated alerts to /alerts
  • Fetches alerts back for confirmation

Testing

• Tested and working locally
• Code follows the style guidelines of this project
• I have performed a self-review of my code
• Code changes documented
• Requested review from >= 2 devs on the team

How to test

1. Place the synthetic nurse notes dataset in the same folder as the pipeline.
2. Run:

python3 guardian_pipeline_final.py

[samjamesbaird-cmyk]

all good, notebook runs smoothly, nice accuracy