fathima86/Oliguric-Acute-kidney-injury-prediction-using-Stacked-LSTM-algorithm ? reverse-engineered prompt

Build me a notebook based project for predicting oliguric acute kidney injury in ICU patients using patient time series data from MIMIC III.

I want one notebook for the stacked LSTM model and another notebook for baseline models like random forest and XGBoost, with the whole flow from loading the extracted CSV files to training, evaluation, and simple result plots. The data should include demographics, labs, vitals, ventilation or vasopressor flags, urine output, creatinine, and the KDIGO based labels. Please also include the preprocessing steps like cleaning, imputation, sequence building for the LSTM, and handling mixed numeric and categorical fields.

Make it easy to run in Jupyter, with clear cells and comments, and include a basic way to inspect model performance. If anything is unclear from the data files or current library versions, look up the latest docs online and make sensible defaults.