A Real-Time Clinical Alerting Pipeline with Graph-Native Temporal Logic
and Multi-Tool Performance Benchmark
Course: DATA 226: Data Warehouses and Pipelines | SJSU
Instructor: Dr. Vishnu Pendyala
Group 5
Sepsis kills 11 million people per year. This pipeline detects sepsis in ICU patients in real time using the official Sepsis-3 clinical rule, then benchmarks Neo4j (graph database) against Snowflake SQL to find the better approach for time-window medical logic.
| Member | Role |
|---|---|
| Trapti Kulshrestha | Ingestion Lead |
| Karthik Pragada | Transformation Lead |
| Poushali Deb Purkayastha | Graph & Benchmark Lead |
| Akanksha Shukla | Orchestration & Delivery Lead — Dagster pipeline, Preset.io, Slack |
- Ingestion: Coupler.io
- Storage: Snowflake (Bronze / Silver / Gold)
- Transformation: dbt
- Graph DB: Neo4j + Cypher
- Orchestration: Dagster Pipeline
- Monitoring: Slack Alerts, Preset.io Dashboard
- Data: MIMIC-IV v3.1 (PhysioNet)