Medical Device Performance Predictive Maintenance Model

Develop a machine learning pipeline using scikit-learn and TensorFlow to predict medical device failure probabilities. The model should: 1) Ingest time-series sensor data from medical equipment, 2) Create feature engineering techniques specific to healthcare technology, 3) Build a predictive maintenance classification model with >90% accuracy, and 4) Generate automated maintenance recommendation reports. Include visualization of prediction confidence intervals and potential cost savings.