MIDAS

MIDAS (Medical Imaging Datasets) is India's first centralised platform for storing, annotating, and accessing medical imaging data — built in collaboration with IISc, ICMR, ABDM, and Niti Aayog. India's fragmented healthcare systems had no shared infrastructure for researchers to obtain quality datasets to train and validate AI models for disease detection.

Six distinct user types needed to interact with the platform — End Users, Platform Admins, Hub Managers, Spoke Coordinators, Annotators, Adjudicators, and Curators — each with different access levels, tasks, and mental models. Designing a single coherent system that served all of them without overwhelming any of them was the core challenge.

Adopted the Carbon Design System as the base and customised it significantly for the medical context. Prioritised transparent workflows and single-click access for non-technical users. Used precise medical terminology in UX writing for the clinical audience. Designed key screens: dashboards per role, data upload interfaces, batch review tools, team management panels, and integrated the Annotator v7 Darwin annotation tool. Moved from mid-fidelity to high-fidelity after stakeholder alignment.

Platform delivered ahead of schedule. Live at midas.iisc.ac.in. Recognised by AIIMS Delhi, ICMR, and IISc, presented at the G20 Summit, and approved by Niti Aayog.