sleep-irregularity

Original Sleep Android SRI Feature Repo

Git clone from this brownhci/sleep-irregularity repo https://github.com/brownhci/sleep-irregularity

• Install Android Studio IDE

  • If on Unix:
    • tar -xvf $zipped_filename_of_android_studio
    • cd android-studo/bin/
    • run ./studio.sh

• Gradle and SDK have to be upgraded to match the configurations written in app > build.gradle

  • If you face Gradle version is not compatible:

    1. cd sleep-irregularity
    2. make a gradle/wrapper dir: mkdir gradle; cd gradle; mkdir wrapper; cd wrapper;
    3. create gradle-wrapper.properties file
    4. Create gradle-wrapper.properties, giving the correct distributionUrl or gradle build version with following information:

    distributionBase=GRADLE_USER_HOME

    distributionPath=wrapper/dists

    distributionUrl=https\://services.gradle.org/distributions/gradle-7.0.2-bin.zip

    zipStoreBase=GRADLE_USER_HOME

    zipStorePath=wrapper/dists

• Configure the gradle project to point to ~/sleep-irregularity/app

• To run java files find a java folder i.e. app/src/main/java hover, right click, click "Run All Tests"
  • If the option to "Run All Test" doesn't show up, try syncing gradle files: Files > Sync Project with Gradle Files

Sleep Regularity Index (SRI)

What is it? It's a measure of sleep regularity by Phillips et al. https://www.nature.com/articles/s41598-017-03171-4 It basically states how consistent your sleep is between two given days on a scale of 100. 100 means you are extremely regular, or in awake state and sleep state consistently at the same times between two days. 0 means you are extremely irregular and that you either were in awake state in day t and complete sleep state in day t+1 or vice versa.

Modified Sleep Regularity Index (mSRI)

This is the average cumulative differences between SRI scores. It is to implicitly show if your regularity fluctuates a lot, or volatility of your regularity.

Sleep Regularity Index - Code Structure

• All testcases are in SleepRegularityIndexTest.java.
• For integration with the pre-existing app, the SRI is called under getSleepIrregularity() under the file SocialJetlagStats.java.

• sriUtils.getSleepIrregularity(records) calculates the average SRI scores
• sriUtils.getSleepIrregularityModified(records) calculates the average MSRI scores

• The SleepRegularityIndexUtil.java file includes the actual implementation of the Sleep Regularity Index.

• It has a process in which it discretizes each 24 hour day to 1440 min representation BitSet
• Let's say there exists BitSet b. If someone is awake at time 14:05, then it's at the 845th minute of the day (14*60 + 5). So b[i] is set to 1. Else, if the person is asleep then the value at b[i] would be 0.
• Now let's say there is day1 BitSet d1 and day2 BitSet d2. A user has "regularity" or "consistency" if d1[i] == d2[i] for all i from 0 to 1440. This means if d1[i] == 0 and d2[i] == 1, there is inconsistent sleep because 0=/=1. There is consistent state if d1[i] ==1 and d2[i] ==1 or if d1[i] == 0 and d2[i] == 0.

SRI Total Calcuation

1. We group all continuous dates and calculate SRI scores. Let's say there is 11-20, 11-21, 11-22, 11-25, 11-26.
2. Then this is grouped into two. Group1: ( 11-20, 11-21, 11-22) and Group 2: ( 11-25, 11-26 ).
3. Since we can only calculate SRI from pairs of days, Group1 will give us 2 SRI scores sri_1, sri_2. Group 2 will give us 1 SRI score sri_3. In total, there are 3 SRI scores.
4. So, we calculate the average as the following: (sri_1 + sri_2 + sri_3)/3, the total_sri_scores/total_num_sri_scores

mSRI Total Calculation

1. Let's say there is 11-20, 11-21, 11-22, 11-25, 11-26, 11-27, 11-29, 11-30.
2. We first calculate SRI scores so we group the dates as follows: Group1: (11-20, 11-21, 11-22), Group 2: ( 11-25, 11-26, 11-27), Group 3: (11-29, 11-30). This results in the following SRI Scores: Group1 : (sri_1, sri_2), Group2:(sri_3, sri_4), Group3: (sri_5).
3. The mSRI score requires 2 consecutive SRI scores. So the mSRI score for Group1 will result in one msri score msri_1 = sri_1 - sri_2 and for Group2 msri_2 = sri_3 - sri_4.
4. Since there is only one SRI score, for GROUP 3, it will not be incorporated into the mSRI. Since there are only 2 mSRIs, the mSRI is then: (msri_1+msri_2)/2.