Quickstart Gudie for Investigators
===================================

Welcome to Datamint! Datamint is a comprehensive platform for medical AI development, with built-in traceability, reproducibility, and audit-ready documentation. This guide will provide a high-level overview of how to use Datamint for your medical AI project. You don't need to have experience developing AI models to use Datamint: we take care of the data management and the AI training, leaving you to focus on the medical research.

Datamint Subscription Plans
~~~~~~~~~~~~~~~~~~~~~~~~~~~
TODO

Projects
~~~~~~~~
To start working with Datamint, create a project. A project in Datamint provides end-to-end support to create an AI tool to perform a specific task. A single project allows you to create multiple AI models which will form parts of a single processing pipeline: for instance, one model which highlights a specific part of a scan, and then a second model which classifies the highlighted section as healthy or diseased.

To learn more about creating a project in Datamint, see :doc:`newprojectreference`.

Uploading Images
~~~~~~~~~~~~~~~~
In Datamint, images (or other medical scans) are kept in the Data Library. Images can be uploaded directly to the Data Library through the Inbox, which allows you to set up automated data pipelines to recieve images from different sources. You can also upload files directly to a project, and they will then also appear in the Data Library. 

Files in the Data Library can be part of more than one project, allowing you to use your data to develop multiple AI tools at the same time. The files can also be tagged and sorted in various ways. When uploading from an existing filesystem, names and pathways are preserved, meaning that Datamint will work seamlessly with whatever existing tools or workflows you have.

To learn more about uploading files to the Data Library, including how to automate the process, see :doc:`inboxreference`. To learn more about how files are stored in the data library, see :doc:`datalibraryreference`.

Annotation Worklists
~~~~~~~~~~~~~~~~~~~~
Datamint allows you to easily coordinate the work of multiple annotators, collating their work in a hassle-free, well-organized way. In order to gather annotations for a project, you must create an annotation worklist. This worklist consists of images you want annotated, and annotation specifications which provide a structure for contributors to create the annotations you need.

To learn more about creating an annotation worklist, see :doc:`annotationworklistreference`. To learn about how to use Datamint's user-friendly annotation tools, see :doc:`annotationtoolsreference`.

Team Management
~~~~~~~~~~~~~~~
Whether you have a team of one, ten, or a hundred, Datamint's robust access control system allows you to delegate tasks while maintaining security and control. You can easily require annotations from specific contributors to be reviewed before acceptance, change what contributors are able to see, and allow other contributors to manage permissions.

To learn more about team management, see :doc:`accesscontrolreference`.

Model Training
~~~~~~~~~~~~~~~
TODO

Validation Studies
~~~~~~~~~~~~~~~~~~
Once you have an AI model trained, it is important to validate its output. Datamint makes this easy with a validation study workflow similar to annotation worklists. After uploading new images that the model wasn't trained on, you can add readers who will rate AI output or analyze the same images for comparison with the model's analysis.

To learn more about validation studies, see :doc:`modeltestingreference`.

Regulatory Compliance
~~~~~~~~~~~~~~~~~~~~~~
TODO