Real-Time Medical Device Data Stream Processing

Develop a Node.js microservice that can process continuous streaming data from multiple medical IoT devices, including heart rate monitors, glucose sensors, and wearable fitness trackers. Implement a scalable architecture using RxJS for reactive data processing, with built-in anomaly detection algorithms that can identify potential health risks in real-time. Include error handling for device disconnections and create a fault-tolerant system that can reconstruct missing data points using machine learning interpolation techniques.