MindLogger use case: NIMH ecological momentary assessment

MATTER Lab spec docs

View project on GitHub

Status

active

Revision history

Name Date Reason For Changes Version
Arno Klein
Jon
2019-10-18 :books: :construction: replace “user” with “informant” in context of user role
:handshake: combine active subject user role and informant user role into informant role
:fire: drop passive subject documentation
:construction: v1.0 draft
Jon 2019-10-18 :pencil: clarify
update roles to separate manager & coordinator
:construction: v1.0 draft
Jon Clucas 2019-10-08 :tractor: begin converting from nested files to flat page :construction: v1.0 draft
Joel Swendsen 2019-09-22, 2019-10-03 answer open questions about EMA design v1.0 draft

Contents

Open questions

  • Are we including the cost of serving these data as a part of the service provided, or will we arrange some kind of pass-through invoicing?
  • Do we want to capture device serial numbers to know when an informant uses multiple physical instances of the same type of device?

Actors

Roles

Managers

One manager sets all parameters for the EMA study. All managers invite informants to appropriate roles (managers, coordinators, reviewers, informants). No data access unless also informant and/or reviewer).

Coordinators

Coordinators invite informants and other coordinators. No data access unless also informant and/or reviewer. Cannot invite managers or reviewers unless also a manager.

Reviewers

Reviewers can see and download data for informants identified with a custom ID.

Informants

Informants self-report through a mobile app or web app according to their schedule. Informants do not see their own custom IDs.

User journeys

Manager

  1. Navigates to admin.mindlogger.org
  2. Selects (default) Child Mind Institute MindLogger production database
  3. Creates an account or signs in
  4. Selects EMA applet or creates it from existing EMA activity set
  5. Sets schedule for EMA applet
  6. Invites informants, reviewers, coordinators, other managers

Manager/coordinator

  1. Receives an email invitation (one time only)
    1. Includes information about applet, identities of managers/coordinators and reviewers
    2. Accepts or declines the invitation from the email OR is directed to accept or decline the email in MindLogger
  2. Navigates to admin.mindlogger.org
  3. Selects (default) Child Mind Institute MindLogger production database
  4. Creates an account or signs in
  5. Selects invitation (one time only)
    1. Steps through information about applet, identities of managers/coordinators and reviewers
    2. Accepts or declines the invitation
  6. Selects EMA applet
  7. Invites informants, other coordinators

Informant

  1. Receives an email invitation (one time only)
    1. Includes information about applet, identities of managers/coordinators and reviewers
    2. Accepts or declines the invitation from the email OR is directed to accept or decline the email in MindLogger (both options are available to informant)
  2. Navigates to mindlogger.org
  3. Downloads mobile app or clicks into web app
  4. Selects (default) Child Mind Institute MindLogger production database
  5. Creates an account or signs in
  6. Selects invitation (one time only)
    1. Steps through information about applet, identities of managers/coordinators and reviewers
    2. Accepts or declines the invitation
  7. Chooses/updates windows during which to be notified
  8. Gets notification (4× daily)
    1. Clicks notification
    2. Begins activity
    3. If activity not completed by a specified amount of time (10, 20, 45 minutes) after notification, gets reminder alarm
  9. Accesses the EMA applet from MindLogger (at any time)
    1. Activities are displayed in scheduled order
    2. Minimum 2 hour blackout period between instances of the same activity
    3. Activities are conditionally available based on schedule
    4. Sees some visual indications of prior completed activities (eg, times completed/missed)
    5. Clicks any available activity
    6. Begins activity
  10. Can leave user group, delete data, or delete account at any time
  11. Can rejoin after leaving user group, can provide data beyond original 2 week commitment

User interface

  • No backtracking

Notifications

  • Conditional on both time and manipulation of device (ie, don’t alert before phone is first handled in the morning)

Response types

Slider

  • Starts blank (no indicator)

Free text

  • out of scope of focus for this study
  • optional activity, never prompted

Audio

  • out of scope of focus for this study
  • optional activity, never prompted

Scheduling

The first manager should decide what the general structure of the study is concerning number of days (eg, 14), number of assessments per day (eg, 4), and the general time span of the assessments (eg, over an 18 hour period, leaving a minimum 6-hour period for sleep). The informant can then decide what the earliest starting time would be for a first assessment (eg, 6am, 8am, etc, depending on their own personal schedule) as well as the latest assessment time (eg 9pm, 10pm, 11pm etc).

The first assessment of the day would be informant-triggered based on the moment they wake up. This will allow us to provide an assessment as close to awakening as possible. This first assessment will be solicited by a notification at the first moment that they check their phone. If the informant stated they wanted to first assessment to occur only after 6am, then they would not receive this notification if they check the time on their phone at 4:30am if they get up to use the bathroom. It would only occur at the first use of the phone after that 6am informant-selected time. This applies only to the first assessment of the day.

Participants will not select the actual times of their assessments within the other time windows (or time epochs) assigned by the manager as this may introduce bias.

We are going to use random assessments withing each of the time windows (with the exception of the first assessment). Up till now, we have used fixed assessments that were randomized across participants (eg, not all assessments occurred at the same fixed time per participant). We can now switch to fully random assessment times within each time window (again with the exception of the first assessment).

We should also fix a minimum two-hour black-out period between assessments. In this example, the informant may wake up at 4:30am, check his or her phone and get no notification to trigger an assessment, then wake up at 6:20am, get a notification and fill out the EMA assessment, then the next one would be alarm-based (or vibration based if the ringer is turned off) at 11:55am, then 5:02pm, then 9:11pm. The next day, the informant may trigger their morning assessment at 6:08am, then receive assessments at 12:18pm, 4:47pm and 10:02pm.

Daily epochs

  • 4 per day
  • informant-specified
    • morning
    • school/work times
      • lunchtime
      • after school/work
    • evening
  • first assessment occurs first time device is manipulated during morning epoch
  • second ‒ fourth assessments occur randomly within informant-specified epochs, at least 2 hours after the previous assessment
  • informants have 20 minutes to complete an assessment
    • for a given assessment notification, informants can snooze the notification once to shift the assessment time forward 15 minute

Duration

The minimum period should be 2 weeks to be compatible with all previously-collected data. Informants will have the option to provide data beyond the initial two-week commitment.

Content

Active collection

We will have one or two questions at the first assessment upon awakening that would cover the (typically night-time) period since the last assessment the day before. For example, since your last assessment last night, how much did you eat or drink during the night, as well as information about what food and beverage intake.

Final set of items forthcoming.

Passive collection

  • Response time
    • started
    • completed
  • Device

Deployment