With Smart IT’s Ensemble ML Model, a Healthcare Provider increased order predictability by 50%.

A leading healthcare provider in the United States was having the following issues with order forecasting and expenditure tracking:

• A mismatch between the Purchase Order (PO) and the actual invoices supplied by hospital partners.

• There is no procedure in place for precise PO forecasting, and the Emergency Operations Center is unable to cope with changes in POs across expenditure categories (EOC).

Working Experts

Smart IT reduced the interruptions to the PO process by constructing an ensemble ML model (XGB + Seasonal ARIMAX) that takes data from current systems and gives an accurate forecast based on the entire PO trend, including weekly/monthly increase and seasonality. Our experts also included the previous three weeks’ data in the model.

• Included the previous three weeks’ data into the model.

• Incorporated the previous three weeks’ data into the model.

• Used casual models to estimate the EOC level share of the expenditure produced.

• Weekly volatility index balanced data sheets through noise component adjustment.

• Increased weekly performance.

Impact We delivered