Cardiva – Demand Forecasting and Logistics Optimisation
Cardiva
Sector: Health
Business Case
Cardiva is a company in the healthcare sector. Among other products, it manufactures single-use packs for operating rooms. Thanks to the digitisation of the demand management process, Cardiva has been able to service its business model and commercialise these packs under a pay-per-use service model.
Objectives
Cardiva set itself the challenge of optimising its order-to-delivery process.
Use case
The challenge was addressed by sequentially solving the following two phases. 1) Forecasting the demand for surgical consumables, in order to anticipate its customers’ needs. Generate biweekly forecasts at the reference centre level using a time series ML algorithm, based on both business data and exogenous variables that impact consumption. 2) Optimising logistics, both in terms of delivery routes and improving stock levels for the end customer. To achieve this, a linear optimisation problem is formulated and resolved at the plant level.
Infrastructure
Cloud
Technology
Machine learning and deep learning
Data
Historical time series of activity by centre: references, stock, consumption, capacities. External factors: Calendar holidays, weather, atypical events.
Resources
A team composed of different profiles: Cardiva: Business and IT managers. Versia: Data analyst, data engineer, data scientist.
Difficulties and learning
Diversity of consumption behaviour at the reference centre level.
KPIs (business impact and metrics of the model)
Consumption levels, number of replenishments, stock per plant. Results obtained for the scope defined in the challenge: Delivery reduction ability: 20%. Ability to free up shelf space: 18%.
Funding
Dataton Euskadi 2023
Collaborators, Partners
VERSIA BAIC
