Irish Hospital Healthcare Analytics

Complete

Analytics platform ingesting Irish hospital patient reviews from Google Places and SerpAPI, combined with government waitlist data (NTPF), enriched with NLP sentiment analysis, and visualised through interactive dashboards.

Technologies

Apache AirflowAWS S3PostgreSQLPythonApache SparkVADER NLPMetabaseDockerGoogle Places APISerpAPI

Problem

Patient feedback was scattered across multiple sources with no unified view of satisfaction trends, wait times, or recurring complaints across Irish hospital groups.

Approach

Multi-source ingestion DAGs in Apache Airflow with S3-backed XCom, PostgreSQL staging via temp tables, and a decoupled NLP pipeline (VADER sentiment) separate from ingestion to allow algorithm iteration without re-calling paid APIs.

Result

Unified analytics platform with sentiment trends, review counts over time, and top keywords by sentiment across Irish hospitals.

Learnings

S3-backed XCom is essential for large payloads; content hashing enables safe re-runs; API response schemas drift — use defensive parsing chains; decoupling ingestion from transformation allows independent iteration.

Relevance

Demonstrates end-to-end data engineering, NLP enrichment, multi-API integration, cloud storage patterns, and production-ready Airflow architecture.