AI-Mind data governance and data management protection framework

In AI-Mind, data initially takes the form of collected data such as participants’ clinical data (e.g., EEG/MEG measurements, neuropsychological testing, etc). This data is then processed and used for developing machine learning (ML)/deep learning (DL) algorithms, thus data later also takes the form of featured data, training, validation, and test datasets and the trained model, plus more. Due to Det Norske Veritas’ (DNV) designation as a Notified Body under the EU medical device regulation (MDR), and thus DNV’s need to maintain impartiality, DNV will not be part of the latter processes nor work with those types of data. As such, D1.4 has been split into two parts:

• D1.4 Part 1: AI-Mind Data Governance Framework – specific for AI-Mind; this details relevant regulations, policies and standards related to data governance, the AI-Mind Data Strategy and Data Governance Policy, Data Transfer Agreement (DTA) and Data Protection Impact Assessment (DPIA). Based on these, a plan for AI-Mind Data Management and the types and roles of different Data, Processes and Actors are outlined.

• D1.4 Part 2: Managing Data in AI/ML Development Projects in Healthcare – generic for healthcare projects; this details the different types of data used in AI/ML projects, and general considerations on how to handle these types of data.

The splitting of this deliverable into two parts enables D1.4 to cover all aspects related to governance and management of the different types of data in AI-Mind, but at two different levels, where Part 1 is specific for AI-Mind, and Part 2 is generic for healthcare projects.

D1.4 Part 1 provides a framework which will enable AI-Mind partners to exercise authority and control over the management of the AI-Mind data. This will enable data to be managed and processed in a compliant and secure manner by partners which ensures quality, and security.

D1.4 Part 2 introduces the complexities associated with AI/ML development projects in healthcare, outlines the types of data in such projects, and provides general considerations about data quality and management for those data.