Case BAS — AI-supported decision-making at the CIZ

Situation

The CIZ assesses whether people are entitled to long-term care — decisions that directly and deeply affect citizens. To support the assessment, the Decision Advisory System (BAS) was developed: an AI-based model that advises the professional in the decision. Exactly the type of use where the bar is highest: an algorithm that touches access to public provisions — under the AI Act the high-risk domain par excellence.

[VERIFY] the above characterisation of BAS and the CIZ; adjust to the system's actual operation and status. [TO COMPLETE] period, formal role, and what may be shared publicly — coordinate with the CIZ if needed.

NIYA's role

[TO COMPLETE] exact role — e.g. development of the model, functional design, process guidance, implementation.

The core of the assignment: a model that supports the professional without replacing the human judgement.

Approach

Human-in-the-loop as a design principle: BAS advises, the professional decides
Explainability first: advice the professional can understand, test and reasonably set aside
Care requirements (GDPR, proper administration, and the instruments now landed in the AI Act and IAMA) translated into concrete design and process requirements
Iterative development with the operational practice at the table

Result

[TO COMPLETE] result in one or two sentences — claim no figures that may not be shared.

What this proves

NIYA has itself built AI-supported decision-making in a high-risk domain — the experience most advisors know only the assessment framework of. Whoever knows such a system from the inside assesses and implements it for you with different eyes.