Overview of directory structure

Description of the directory structure of a typical mhealth storage compatible dataset

- root
 - [Participant ID 1]
     - MasterSynced
     - OriginalRaw
     - Derived
         - [SetName]
     - Meta
         - location_mapping.csv
         - subject.csv
         - sessions.csv
 - [Participant ID 2]
     - MasterSynced
     - OriginalRaw
     - Derived
         - location_mapping.csv
         - subject.csv
         - sessions.csv
         - [SetName]
 - MetaCrossParticipants
     - location_mapping.csv
     - sessions.csv
     - subjects.csv
 - DerivedCrossParticipants
     - [SetName]

Participant ID

1. Allowed characters: alphabets and digits.

2. Globally Unique.

3. Used as directory name.

MasterSynced

1. Synchronized sensor, feature, event, annotation files, stored in mhealth storage format.

2. Files are divided hourly, in folder MasterSynced/YYYY/MM/DD/HH/

OriginalRaw

1. Original files stored in manufacturer's format.

2. Arbitrary directory structure.

3. Recommended to divide sub directories by the type of sensor or data.

Meta

1. Store meta files that are additional information to files in MasterSyncedfolder.

2. Meta files should be kept unchanged all the time.

Derived

1. Store data files derived or merged from files in MasterSynced folder. Such as feature files, class files, model files, ema summary files.

2. SetName is used to specify any common properties of a set of derived files. Popular names could be AllTime,LabSession,Weekend, FebruaryMorning or the combination between them.

MetaCrossParticipants

1. Store meta information that is aggregated from files in each participant's Meta folders.

2. Meta information is the unchanged information for a dataset.

DerivedCrossParticipants

1. Store data files derived and merged from files in each participant's MasterSynced or Derived folders.

2. SetName is used to specify the time or participant characte of a set of derived files. For example, AllTimeMale, LabSessionAdults and so on.