At a glance

One of the largest states in the US was in search of a predictive model to help them manage the demand and supply of PPE kits during the Covid-19 pandemic. Eviden collaborated with the state’s Emergency Services and an American multinational technology company to create 55 models for demand forecasting of different PPE kits — a crucial prediction in a medical emergency.

Outcomes

  • Accuracy in prediction of the PPE kit inventory
  • Optimal inventory levels met

The client

The Emergency Services of one of the largest states in the US is responsible for sourcing personal protective equipment (PPE) kits for the state and all its healthcare facilities. It is responsible for planning, procuring and distributing PPE equipment. This also includes managing any supply disruptions so that it can meet the demand placed by these facilities.

Decoding demand in a medical emergency

Amid the COVID-19 pandemic and the medical emergency that ensued, this US state needed an accurate forecast of the demand and supply of PPE kits for their facilities. This had been impacted by different dynamic variables that the 14-day sliding average of demand had proven to be an unreliable predictor of actual future demand. The daily changes in the statistics and hospital data, inconsistent order of county data, lack of data around PPE kit consumption and insufficient historical data were some of the challenges faced by the state.

Predict, protect, preserve

Eviden partnered with the US state to create multiple machine learning and deep learning (ML/DL) models for different counties, and product family combination to overcome data related challenges.

As a part of this solution, Eviden completed complex feature engineering operations and created feature stores to fast track the model development process. With a robust MLOps pipeline, Eviden delivered 55 models that were deployed and executed after training the end users. These models utilize historical data from the states’ orders and hospital data.

In combination, the PPE forecasting model utilizes outputs from another ML/DL model — the Cloud AI COVID19 — that was developed by an American multinational technology company focusing on artificial intelligence, online advertising, search engine technology, cloud computing, computer software. This DL model is designed to predict the number of cases, hospitalizations, and deaths.

The key variables from the PPE forecast model is a time-based demand forecast at a county level, which may be used as a guidance for the state’s purchase decisions and ensure an optimum level of inventory.

AI to the rescue

This AI solution has delivered a 95% accuracy in the prediction of PPE kit demands and optimized inventory levels at a Optimize Warranty Reserve level.

AI is a critical tool in a medical emergency such as the pandemic, ensuring transparency, tracking and availability of supplies in a time of crisis.

Why Eviden

As your trusted IT partner, Eviden leverages its proven expertise in Machine Learning solutions. Our highly skilled ML engineers use advanced forecasting algorithms to optimize resource planning, ensuring efficiency and minimizing shortages.

Related resources

Client story

Computer vision for automated inspection of products delivered from the assembly line

One of the largest manufacturing conglomerates from Germany uses Computer vision for automated inspection of industrial equipment.

Client story

Connected autonomous vehicle

Eviden assisted a leading US automaker in crafting a cutting-edge Connected Car Platform for innovative R&D.

Client story

Intelligent conversational agent

Eviden designed a Intelligent Conversational Bot for one of postal service provider in South Europe.

Start your story with Eviden

Upgrade your expectations! Discover excellence tailored to your needs. Experience extraordinary service, innovative solutions, and unparalleled support. Join us today for a remarkable journey!