Real-Time Patient Vital Signs Anomaly Detection System

Design a Python microservice using Flask and scikit-learn that processes streaming medical sensor data in real-time, detecting potential health anomalies. Create machine learning models to identify statistically significant deviations in vital signs like heart rate, blood pressure, and oxygen saturation. Implement a multi-layered alert system that can trigger different notification levels based on detected risk severity, with integration capabilities for hospital monitoring systems.