Introduction

Envision Portal is a web platform for managing, standardizing, and sharing biomedical research datasets. It is designed for research teams working with eye imaging data and other clinical datasets who need a structured way to document, version, and publish their work.

The platform is developed within the Eye ACT project to address common barriers in ophthalmic data reuse: fragmented sharing practices, missing metadata, and inconsistent file formats. Envision Portal focuses on making datasets more Findable, Accessible, Interoperable, and Reusable (FAIR) and better prepared for downstream AI workflows.

What Envision Portal Does

The platform covers the full lifecycle of a research dataset:

• Create datasets with rich metadata following established standards (Dublin Core, DataCite, FAIR principles)
• Document datasets with clinical trial study information and data quality documentation (Healthsheet)
• Upload files to cloud storage (Azure Data Lake) with separate environments for drafts and published data
• Publish datasets through a guided multi-step review workflow that generates a README and changelog
• Control access with role-based team permissions, public access requests, and controlled access workflows with data use agreements
• Discover publicly available datasets and register external datasets from other repositories

The long-term scope also includes automated indexing of eye imaging datasets from external repositories and advanced natural-language discovery features.

Who It Is For

Envision Portal is intended for:

• Researchers who create and manage datasets and need to document and share them
• Data stewards who review access requests and manage team permissions
• Collaborators who need read or edit access to a dataset
• External users who want to discover and request access to published datasets

Key Concepts

Datasets

A dataset is the central object in the platform. Each dataset has a canonical ID that persists across versions, a current version number, and a status of either draft or published.

Metadata

Each dataset supports three layers of metadata:

Access Levels

Datasets can be accessed at different levels:

• Public access allows anyone to request and download the dataset
• Controlled access requires a formal request, a signed data use agreement, and approval from the dataset owner

The product roadmap also tracks support for additional access patterns (for example, restricted or policy-based access tiers) as governance requirements evolve.

Team Roles

Every dataset has a team of members with one of four roles: owner, admin, editor, or viewer. Roles determine what actions each member can take on the dataset.

Supported Data Standards

Envision Portal guidance and validation workflows are aligned with standards used for interoperability and reuse:

• CDS (Clinical Dataset Structure) for consistent multi-modal dataset organization
• DICOM for eye imaging files wherever possible
• OMOP CDM-aligned tables for associated clinical and observational data
• FAIR metadata practices to improve discoverability and reuse

These standards are expected to evolve with the community, and documentation will be updated accordingly.

Development Principles

The platform is developed with repository trust and sustainability in mind, including:

• Transparent public documentation and open-source development
• Persistent identifiers (DOIs) and reproducible versioning
• User-focused workflows for both contributors and data consumers
• Secure, maintainable cloud infrastructure and policy-aware access controls

Roadmap Snapshot

The Envision Portal roadmap spans 2024-2029, with phased milestones:

• Initial architecture and standards definition
• Early launch with hosted and externally indexed datasets
• Contributor-facing automation for standardization and PHI checks
• Advanced discovery, APIs, and controlled-access workflows
• Sustainability, security hardening, and community-led growth