Using Kibana

Avicenna Kibana Integration

Kibana is an open-source data visualization tool which can create different graphs and charts from large amounts of data. Kibana allows you to explore the data, create visualizations such as bar chart, line charts, scatter plots, maps, and many more. You can further combine these visualizations to create interactive dashboards.

In this document, we describe how you can use Kibana to explore and visualize your Avicenna study data. This document does not intend to teach you how Kibana, Elasticsearch, or Lucine works. There are already many online resources and training videos for these technologies. We suggest you review them to better understand how you can use Kibana for your work.

For each study you create in Avicenna, a set of data tables are created in a data storage system called Elasticsearch. These tables are also called index or index pattern. Kibana allows you to access these data tables, query them, read the data, and visualize them. To access the Kibana, go to the Researcher Dashboard, and from the left-side menu click on Kibana. This will take you to a page similar to the image below:

There are 4 important sections in the above image:

Number 1 lists the data tables for your study. Your study does not necessarily have data in all these tables, depending on which data sources and which activities you have added to your study.

Number 2 shows the time range over which you are querying the data. By default, the time range just shows the data for the past 15 minutes, so there is a chance that you do not see any data. You can increase this time range to load more data.

Number 3 lists the data fields that are available in the current data table. This list is different depending on the data table that you have selected. For example, for GPS it includes the location and the speed of the movement, while for Pedometer it includes the number of steps taken.

Number 4 allows you to filter your data. You can put filter on any data field that exist in the current data table. For example, assuming that you are looking at the Pedometer data table, you can filter data for those who have taken 100 steps or more.

Avicenna does not store all data sources in the Elasticsearch. At the moment, only the following data sources are stored in the Elasticsearch and therefore are accessible through Kibana:

• GPS (gps index)
• Wi-Fi (wifi index)
• Pedometer (pedometer index)
• Motion-based Activity Recognition (mb_activity_rec index)
• Battery (battery index)
• Screen State (screen_state index)
• Bluetooth (bluetooth index)
• Bluetooth Beacons (beacon index)
• App Usage Statistics (app_uage index)
• Call & SMS logs (telephonycomms index)
• Participant Audit Logs (participant_history index)
• Survey Responses (survey_responses index)
• Stroop Responses (stroop_responses index)
• Time Use Diary Responses (time_use_responses index)

Below you can read the details for each of these data sources, including available fields and their types.

Avicenna clusters host the services for Kibana and Elasticsearch. Therefore, your study data does not leave Avicenna servers for any of the procedures described here.

Available Data Sources

This section describes the available data sources and the fields each data source have in Kibana. The name of the field is necessary to access the field's data in Kibana. The field's type specifies the type of the data being stored in it. You can refer to Elasticsearch's field datatypes for more details about each type. You also can check the Data Sources section for detailed definition of each field.

Common Fields

The following data fields are available in each of the data sources described below:

Study ID
Filed Name	study_id
Type	Integer
Description	The ID of the study this record belongs to.

Participant ID
Filed Name	user_id
Type	Integer
Description	The ID of the participant who provided this record.

Device ID
Filed Name	device_id
Type	Keyword
Description	The ID of the device which provided this record. Each participant can own multiple devices during the course of the study, and each device will have a unique ID. Avicenna uses this ID to tag all records coming from the same device. The ID remains the same even when the user uninstalls and reinstalls the Avicenna app on their phone.

Record Time
Filed Name	record_time
Type	Date
Description	The time which this record was captured. For survey responses, record_time for all responses in the same session are identical, and it represents the time when the user has pressed Submit button (or equivalent) to finish responding to the survey.

Relative Record Time
Filed Name	rel_record_time
Type	Date
Description	The number of milliseconds between the participant's participation period start time, and the time this record was captured. This field is particularly useful for studies with rolling enrollment, where each participant starts the study at a potentially different date. Therefore, 0 indicates the record was captured right at the start time, 1 indicates the record was captured 1 ms after the start time, and so on. Note that this field is marked as Date, therefore Avicenna will show the field as milliseconds passed Unix epoch (Jan. 1st, 1970). If you plan to query the data based on this field, you need to set the time range based on this date.

Survey Responses

For survey responses, Avicenna stores each response to a given question as a separate record. Therefore, a given survey session can contain multiple records. For example, assume your survey contains 5 questions, from question ID 1 to 5. Every time a participant responds to your survey, 5 new records will be added to this index, one for each question (assuming the participant has responded to all questions).

Also, note that not each record contains all the fields specified here. If a given record does not have a given field, it means the field was not relevant for that record. For example, if a survey response is of type text, the record will contain answer_content, but it will not contain answer_url.

Index name: survey_responses

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Survey ID	survey_id	Integer
Question Set ID	questionset_id	Integer
Response Duration	duration	integer	In minutes.
Scheduled Time	scheduled_time	Date	For Time- and Proximity-Triggered sessions, this shows the time the survey was automatically triggered. For User Triggered sessions, this shows the time the survey was started by the participant.
Prompt Time	prompt_time	Date	Same as Scheduled Time.
Response Time	resp_time	Date	The time this response was provided.
Iteration	iteration	Integer
Loop Count	loop_count	Integer
Question ID	q_id	Integer
Question Content	q_content	Text
Question Type	q_type	Keyword
Answer ID	answer_id	Integer
Answer Content	answer_content	Text
Answer URL	answer_url	Keyword
Location	location	Geo Point
Location Accuracy	location_accu	Double
Location Speed	location_speed	Double
Preferred Unit	pref_unit	Keyword
Selector	selector	Keyword

GPS

Index name: gps

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Provider	provider	Keyword
Satellite Time	satellite_time	Date
Location	location	Geo Point
Speed	speed	Double
Accuracy	accu	Double
Altitude	alt	Double
Bearing	bearing	Double

Wi-Fi

Index name: wifi

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
SSID	ssid	Keyword
BSSID	bssid	Keyword
Access Point Capabilities	capabilities	Keyword
Frequency	freq	Integer
Level	level	Integer

Pedometer

Index name: pedometer

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Steps	steps	Integer
Accuracy	accu	Double
Distance	distance	Double
Average Active Pace	avg_active_pace	Double
Current Cadence	cur_cadence	Double
Current Pace	cur_pace	Double
Duration	duration	Integer
Floor Ascended	floor_ascended	Double
Floor Descended	floor_descended	Double

Motion-based Activity Recognition

Index name: mb_activity_rec

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Activity Type	activity_type	Integer
Confidence Level	confidence_level	Integer

Battery

Index name: battery

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Level	level	Integer
Scale	scale	Integer
Status	status	Integer
Plugged	plugged	Integer
Temperature	temperature	Integer
Voltage	voltage	Integer

Screen State

Index name: screen_state

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
Screen State	state	Boolean
End Time	end_time	Date

Bluetooth

Index name: bluetooth

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
MAC Address	mac	Keyword
Device Name	dev_name	Keyword
Device Class	dev_class	Keyword
RSSI	rssi	Integer

Bluetooth Beacons

Index name: beacon

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
MAC Address	mac	Keyword
Device Name	dev_name	Keyword
Device Class	dev_class	Integer
Payload	payload	Long
Team ID	team_id	Integer
Role ID	role_id	Integer
Subject ID	subject_id	Integer
RSSI	rssi	Integer

App Usage Statistics

Index name: app_usage

Index fields:

Name	Field Name	Type	Description
Participant ID	user_id	Integer
Device ID	device_id	Keyword
Record Time	record_time	Date
Relative Record Time	rel_record_time	Date
App Name	app_name	Keyword
Start Time	start_time	Date
End Time	end_time	Date
Last Used	last_used	Date
Foreground Time	foreground_time_ms	Integer	In milliseconds.