MediBRAIN develops an AI-powered decision support system for the personalised management of diabetes. Its aim is to turn the volume of data generated daily by a person living with diabetes — glucose readings, personal health record entries, wearable device data — into information that can genuinely be acted upon in everyday life.

At the core of the system are machine learning models that predict glucose levels up to thirty minutes ahead, personalised to each user. Algorithm development and evaluation were carried out using twelve anonymised digital twins, while interoperability is secured through FHIR R4 resources and modular APIs that allow the system to connect with the wider health ecosystem. Results are presented to the user through the MedInfoBook platform.

Privacy and GDPR compliance were design principles from the outset of the project. The full set of deliverables — from market research and requirements analysis through to technical verification, pilot evaluation and commercial exploitation — is published openly on the project website.

Scope note: The current implementation is a proof of concept and pilot system based on simulated data and digital twins. It does not constitute a clinical trial, it is not a certified medical device, and it is not intended for clinical decision-making.