Wearable Device Time Series Health Anomaly Detection

Create an advanced anomaly detection system for continuous health monitoring using time series data from wearable fitness devices. Develop a Python script using Prophet for forecasting and Numenta's NAB (Numenta Anomaly Benchmark) algorithm to identify statistically significant deviations in physiological metrics like heart rate, sleep patterns, and activity levels. The solution must handle real-time streaming data, generate interpretable alerts, and provide a machine learning model that can adapt to individual patient baselines.