General principles

lang: en-US title: General principles description: Description of the general principles of the CDS

---

Definitions

• The keywords "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in RFC2119.
• Clinical dataset: A set of data files and associated metadata files resulting from a clinical research study.
• Datatype: In a clinical research study, multiple modalities of data are collected. A datatype designate one modality or a group of modalities. We postulate that data from multiple modalities can be characterized as being of a single datatype if at least one of these is applicable:
• There exists an established standard structure for organizing data from these modalities together
• The modalities were collected through a shared method (instrument, device, etc.)
• The modalities cannot be interpreted separately
• The modalities are typically collected together

General specifications

The CDS specifies the following:

1. No data file must be present at the root level.
2. The data files must be organized into one directory per datatype at the root level as per the specification provided here.
3. Within each datatype-specific directory, data and metadata files must be organized according to an existing standard for that datatype, if available, or following the CDS-suggested structure, as explained here.

4. The following metadata files must be included at the root level:

   • README.md
   • LICENSE.txt
   • CHANGELOG.md
   • healthsheet.md
   • study_description.json
   • dataset_description.json
   • participants.tsv and participants.json
   • dataset_structure_description.json

5. No empty directories must be included anywhere in the dataset.

Illustration of the root level of a dataset structured following the CDS.