Real-Time Medical Time Series Database Optimization

Architect a high-performance time series database for continuous patient monitoring data using TimescaleDB with Python. Develop an optimized schema that can ingest high-frequency medical sensor data (heart rate, blood pressure, oxygen levels) with sub-millisecond write latency and complex analytical query support. Implement automatic data retention policies, hypertable partitioning, and compression strategies to manage potentially millions of patient readings per hour.