top of page
Virtual Colleague
Unlocking internal & external knowledge for elderly care professionals

Stage:
Roll-out phase
Country:
The Netherlands
Partners:
Tante Louise, Mijzo, Care Innovation Center, Interreg, CrossCare & Nederlandse Ministerie van Economische Zaken
Ask your Virtual Colleague
Leveraging customised Large Language Model to provide elderly care staff with timely and pertinent information through a Virtual Colleague interface.
Existential need for efficiency gains
The European healthcare sector grapples with a prolonged shortage of personnel exacerbated by dual aging populations and increasingly complex care needs. These evolutions will put immense pressure on the quality and accessibility of healthcare in the coming decades.
At the operational level care workers spend about 10-15% of their working time looking for relevant answers to practical and medical questions. The data is available, but not readily accessible.
Artificial intelligence (AI) holds promising solutions. In the "My Virtual Colleague" project, sophisticated AI (Large Language Models) is employed to equip healthcare workers and learners with relevant information irrespective of time and place.
Unlocking relevant data
With the "Virtual Colleague" we built a custom Large Language Model to equip healthcare workers with relevant information irrespective of time and place. The model is trained on both internal company data and relevant public data to answer the most common questions asked by the care workers.
In the case of Tante Louise, the Virtual Colleague is trained to answer both practical questions and searches related to medical protocols. Mijzo leverages the solution to support students throughout their training period.
Benefits
The Virtual Colleague brings three key benefits:
Immediate Support - Prompt handling of staff queries, avoiding delays in the workflow
Increased Productivity - Less time spent researching and answering routine inquiries
Quality of care - Better access to relevant data increases the quality of services and reduces mistakes.

bottom of page