About GSMS

GSMS, founded in 1986, is the leading provider of generic pharmaceuticals to the US Federal Government with the primary focus on serving the Department of Défense (DOD) and Veteran Administration (VA) through its private label. It operates within a niche market serving as value-add intermediary between generic pharmaceuticals manufacturers and the VA and DOD. GSMS offers a large portfolio of prescription products across a broad range of dosage forms. All GSMS products are labelled in accordance with FDA guidelines for standardization and consistency.

Challenge that led GSMS to Have a Robust Demand Forecasting Model

Accurate demand forecasting is the foundation for any supply chain and is extremely critical for the pharmaceutical industry in particular, where seamless supply of the right medications in the right dosages on time is crucial for positive health outcomes. GSMS aims to stay competitive in the generics market, which led them to the challenge of accurately forecasting demand. They wanted to leverage predictive insights in their existing forecast model to ensure a seamless supply chain and timely delivery of medications, while minimizing the risks associated with inventory management.

GSMS will leverage Finarb’s expertise, a leading provider of AI and advanced analytics solutions in the Healthcare industry. As a part of the partnership, Finarb will make use of GSMS’s proprietary data and other publicly available information, extract relevant features from the data by using advanced techniques, build predictive models to address the problem statement and help facilitate effective decision-making so that GSMS remains at the forefront of innovation and technology.

Data Requirements and Predictive Modelling

In our forecasting methodology, we employed two sophisticated techniques: Classic Time Series and ARIMAX. The variables we considered include economic indicators, competitor activities, seasonal events, disease incidence, veteran demographics etc.

By integrating these two techniques, we derived a robust and comprehensive forecast. To train these models we used market sales data including sales volume, customer demographics, product-wise sales, sales trend etc. We also analysed historical sales data, market trends, seasonal demand patterns, and sector forecasts to forecast market demand. Patterns in daily transactions were learned from sales chargeback data which included transaction details, reasons for dispute, customer feedback etc. In addition, we also considered the competitor pricing data, regulatory changes, and market trends to arrive at our forecasts.

Outcome & Benefits

Model performance: Our solution achieved an overall Mean Absolute Percentage Error (MAPE) of 0.295 compared to the existing MAPE of 2.94, a 90% reduction.

This will enable GSMS in better demand planning, reducing inventory risks and reaffirms its commitment to seamlessly serve the critical needs of the US Federal Government.