Healthcare Cost Prediction Machine Learning Model

Develop a predictive regression model using Python that estimates patient treatment costs with 90% accuracy. Utilize a dataset of 50,000+ medical records, incorporating features like diagnosis codes, patient demographics, treatment history, and regional healthcare pricing. Implement cross-validation with scikit-learn, feature engineering with advanced techniques like polynomial features, and create a Flask API endpoint for real-time cost predictions. Include comprehensive model interpretability using SHAP values.