AI Supply Chain Optimization Solution for Platelets

Summary:

An AI supply chain optimization solution for platelets enabled the NHS to achieve cost savings, reduce waste, and provide more efficient patient care.

Client: 

The National Health Service (NHS) in the UK manages healthcare systems, serving millions of patients annually. 

Problem Statement: 

Efficiently managing the supply chain of critical medical supplies, such as platelets, is vital to minimize waste, reduce costs, and enhance patient care. Platelets, due to their short shelf life of five to seven days, require precise demand forecasting, inventory management, and logistics coordination. Challenges like overstocking, shortages, or expired stock are common, posing both a financial burden and a risk to patient health.

Results: 

• Achieved a 54% reduction in expired platelets.
• Completely eliminated the need for costly ad hoc transport.
• Reductions in expiries while still keeping the “In Full” delivery rate at the same high level.

AI Solution Overview:

Kortical deployed an end-to-end AI-driven optimization solution that leveraged ML, data integration, and advanced predictive analytics to transform the NHS platelet supply chain. 

Kortical initially tested its solution across select NHS facilities to assess performance, collect feedback, and refine the predictive models.

The initial approach for implementing ML involved utilizing NHS data to forecast demand for 40 different blood products across 15 distribution hubs. Each day, the ML model generates predictions on the quantity of platelets that hospitals will order for each blood product and every stock-holding unit.

By integrating logistics data, Kortical AI’s system identified optimal distribution routes and schedules to minimize delays and ensure timely delivery, thus maximizing platelet shelf life at points of care.

The data was subsequently input into the Kortical AutoML platform, where the patented AI as a Service automatically created thousands of machine learning models using a variety of algorithms, such as Deep Neural Networks and Extreme Gradient Boosting. The platform enhanced these models with automated feature engineering and data cleaning processes, ultimately producing the highest-performing models tailored to the business problem at hand.

For this specific demand prediction challenge, the top-performing model was XGBoost.

References: 

1. AI supply chain optimisation for platelets to reduce costs. https://kortical.com/case-studies/ai-supply-chain-blood-healthcare-nhs

Industry: Healthcare

Vendor: Kortical

Client: The National Health Service (NHS)