Intro

Prospitalia GmbH is one of the leading purchasing service providers in the German healthcare market and supports over 1,300 hospitals, hospital pharmacies and numerous other healthcare facilities with innovative purchasing solutions and services. The aim is to increase efficiency, reduce procurement costs and ensure the quality of patient care in the long term.

With the company’s own “Prospitalia Cockpit” portal, Prospitalia provides its contract clinics with a powerful database that offers quick and uncomplicated access to relevant article information.

The precise classification and allocation of products ensures high data quality, optimizes process efficiency and creates a reliable basis for strategic purchasing decisions.

THe Challenge

Time-consuming processes and data inconsistencies as key challenges

An essential part of Prospitalia’s value creation process is the preparation of supplier article master data so that it can be easily found and evaluated in the “Prospitalia Cockpit”. The focus here is on the precise classification of articles and the storage of negotiated conditions. This addresses an industry-wide problem: Only around 60-70% of item master data is fully classified, i.e. structured and standardized. These gaps not only make operational purchasing more difficult, but also block strategic measures such as portfolio optimization or data-based analyses, which are essential for efficient cost management.

Previously, this data preparation was largely done manually. However, the constantly growing number of articles and the dependence on expert knowledge led to an increasing backlog.

Item matching was also a challenge: In order to always provide the contractors with up-to-date information, possible replacement products and the best possible negotiated conditions, the items requested by the clinics had to be compared with the item data in the Prospitalia database.

What seemed simple at first turned out to be extremely complex in practice. Different data standards and article descriptions in the materials management systems of the suppliers and clinics made manual reconciliation laborious and time-consuming. The resulting processing backlog ultimately impaired the user-friendliness of the platform.

Solution

Artificial intelligence as the key

To solve these challenges sustainably, Prospitalia implemented the Product Data OS from entitys.io. This SaaS solution uses integrated machine learning (ML) functions to intelligently automate both the classification and allocation of product data.

Classification-as-a-Service

A specially trained ML model analyzes the product descriptions and automatically suggests suitable ECLASS categories. Instead of a manual search, users can now simply select the correct classification from a suggested list. The result: a significantly shortened process, reduced error rates and less reliance on specific expert knowledge.

Clinic MaWi article master mapping

In addition, a solution was developed that compares products from different sources. With the help of ML algorithms and natural language processing (NLP) technologies, the best matches are identified even in the case of incomplete or widely differing descriptions. Clinics can thus reliably access the requested products in the central database and directly retrieve relevant information for purchasing decisions.

Results