Several technical factors are important to ensure that the AI system functions properly within the organization's infrastructure. It must be well integrated into the workflow, fit the environment in which it will be used, and be easy to use. It is important to remain open to making significant changes and thinking innovatively, such as prioritizing the establishment of efficient infrastructure to ensure good workflow, rather than merely improving existing but limited systems.
Data management, quality and compatibility
Those responsible for the organisation's information governance are important in the mapping process and must be involved. The organisation should plan for holistic information management. The guide Getting your own house in order from the Norwegian Digitalisation Agency
can be useful in both mapping and planning [101]. The organisation should have assessed the requirements that will apply to the processing of data in the AI systems and whether the organisation's data meets these requirements. This applies, for example, to requirements for data quality, data format and structure, encryption, compression, data labelling and storage methods.
Integration of the AI system
When assessing whether an AI system is suitable for the organisation, it must be clarified which interfaces the organisation uses for integration with local processing systems and electronic health record systems, as well as which open standards these interfaces comply with. Effective data handling and interoperability between systems are essential for ensuring efficient processes.
Storage, processing power and infrastructure
The AI system must be evaluated in light of any technical infrastructure requirements the organization may have, for example, regarding processing capabilities, storage, network, monitoring tools, security, and software licenses. It should be assessed whether changes to the technical infrastructure are needed for testing, implementation, and ongoing operations.
Evaluate test environments
Test environments may be necessary when procuring AI systems, as they enable the organization to conduct quality assurance in a controlled setting. Requirements for test environments during the various phases should be clearly defined, including the technical infrastructure needed, whether the supplier is expected to provide (parts of) the test environment, and the need to test integration with existing systems.
Both initial and ongoing operational costs associated with the test environments should be estimated. The supplier must specify whether additional costs will be incurred for access to their own test environments, or for using the AI system within the organization's test environments — for example, additional licensing fees.