AI Use Case for Rebate and Contract Optimization


Enabled quick evaluation of multiple datasets

Built a predictive model for scoring based on market share by plans, payers, and PBM’s formularies

Real-time, dynamic visualization of the impact and ROI for each Payer, brand, contracted, or non-contracted market share, and expected ROI as the basis for decision making


Identify opportunities for contract optimization for a portfolio of products

Analyze revenue and rebate projections based on contracted access vs. noon-contracted access

Calculate expected ROI from Contract for each brand within the portfolio and by each Payer


The primary objective for this project was to ensure that Gross To Net (GTN) projections are accurate and provide GTN optimization
opportunities. Once accomplished, this will create the opportunity to Identify opportunities for GTN Optimization for a given Portfolio. Revenue Projection based on Contracted Access vs. Non-Contracted Access


Leverage available data streams at various time intervals, including market and brand coverage. Data from
these sources were accessed and combined for analysis:

IMS Plantrak Sales at a weekly physician, plan, and brad level

MMIT Formulary Access monthly data at a plan and brand level

LAAD half-yearly data at the physician, plan brand, and semester level

Copay and FIA data at a Plan, brand, and semester level

TN data at the yearly, brand level

Data integration analysis uncovered data anomalies and outliers. Applied univariate analysis, descriptive and correlation analysis, bivariate exploratory data analysis (EDA) and time series EDA.

Machine Learning Models (light GBM at plan level) were applied to predict likely share at different formulary coverage at the Plan level and created ROI & Break-Even Analysis. Statistical evaluation of predictive models identified the best models. Analysis of “what if” scenarios to assess likely profitability of various contracted access at the Payor level for an essential brand. This analysis enabled future GTN projections for the portfolio.