Real-Time Medical Device Data Normalization Pipeline

Create a complex SQL ETL process that can ingest streaming medical device data from multiple sources (ECG monitors, insulin pumps, wearable sensors) with varying data formats. Develop a robust normalization strategy that handles timestamp discrepancies, converts multiple unit measurements, detects and quarantines anomalous readings, and creates a unified schema for clinical research. Include error handling for incomplete datasets and implement a sliding window analytics approach.