Quantified Self Experiments / Tracking - Sleep

Why to track?

There is a lot of good books revealing importance of sleep: Why we sleep by Matthew Walker, Circadian Code by Satchin Panda - based on scientific evidence. In short, sleep is evolutionary mechanism for body recovery, cleanup, growth, development, learning, memory preserving etc.

Since we know that sleep quantity and quality degrades during aging, at the same time chronic sleep disorders arising. Degraded sleep impacts body homeostasis, metabolism, physical and cognitive functioning and results in lowering our energy, productivity and quality.

What can we do to combat these issues and prolong our healthspan?

  1. Measure sleep quality and quantity
  2. Measure how different factors affecting sleep
  3. Optimize these factors and check if sleep outcomes improved
  4. Repeat until there is no room for valuable improvement

Analyzing sleep data can answer important questions:

What to track?

Gold standard for sleep measuring is a PSG study in lab. Since it very expensive, time-consuming, hard to implement in long term at home and not convenient for use - there are some alternative ways to measure. What metrics important enough to be objectively measured?

Some additional metrics can complement sleep data, including subjective ones

How do i track?

There is a several consumer devices on the market which track some of metrics from previous block and i've tried: Oura ring, Withings Sleep, Dreem 2, Fitbit Charge 4, Garmin Vivosmart 4, Polar Vantage V2, Wellue SPO2. Since i havent done PSG, i'm relying on Dreem 2 as my gold standard [1] [2].

Here list of accurate sleep metrics compared to Dreem 2:

I was totally dissatisfied with accuracy of wrist / finger wearables. They can't be alternative to EEG in measuring sleep quality. After realizing that, i've stick to this setup:

Since Fitbit & Oura are pretty bad at detecting AWAKE, SOL, DEEP, REM, LIGHT their recovery / sleep scores are useless. They cant be used for sleep quality assessment and very limited for sleep quantity assessment, because AWAKE detection is inaccurate and they cant detect TST, which is main sleep quantity measure.

It seems that in case of running self experiment only Dreem 2 can accurate measure important outcomes. I'm not using Audio/Video recordings because of their dedicated use in diagnostic of specific disorders and lack of automation. Enviromental factors can be measured with devices like Atmotube Pro or similar.

Sleep Diaries and Questionnaires seems to be presented in science for a long time. I'll list only few, which i'm using right now.
Weekly sleep quality score, for long time trends:

There also some mood issues connected to sleep: Depression, Anxiety, Excess Fatigue, Stress. I measuring them by Center for Epidemiologic Studies Depression Scale (CES-D), General Anxiety Disorder (GAD-7), Perceived Stress Scale (PSS), Fatigue Severity Scale (FSS)

For a some time, for daily resolution, i've rated my subjective sleep quality in the morning on a 7-point Likert scale: Very poor, Poor,  Less than Avg,  Average,  Good,  Excellent, Perfect. After some time i found this single rating wasnt sensitive enough, with most of time being Average.

After reading scientific papers about daily subjective sleep assessment [3] [4] [5] [6] i've made my personal morning sleep questionnaire based on objective and subjective measures from Pittsburgh Sleep Dairy:

  1. How much time you spent in bed? (TIB)
  2. How long it took to fall asleep? (SOL)
  3. How much time awake after sleep onset? (WASO)
  4. How many times you wake up during the night?
  5. Bad sleep because awake in middle of night or early morning and trouble sleep back? (Yes/No)
  6. Rate of heaviness in head (7-point LikeRT)
  7. Rating of Cheerfulness after waking up (7-point LikeRT)
  8. Rating of Sleep Recovery after waking up (7-point LikeRT)
  9. Overall Sleep Quality rating (7-point LikeRT)

Answering these 9 questions every morning taking less than 1 minute, but requires some discipline. Telegram bot sends me questions and i need to tap button to answer - very easy, fast and convinient, results automatically saved into database.

Since PSQI scores set of similar questions into final index, i'm using same approach and in future will validate this questionaire against Dreem 2 objective measures. My point that final index should have some correlation with metrics from Dreem 2. If it's true, there will be a greater chance to find effect in my future experiments by using only morning questionnaire. I think some of QSers failed to find effect on sleep by their interventions because were using rough low-sensitive single item scale.

I'm also using PANAS-20 every evening to assess Positive and Negative Affect

Since it have scales in some way related to sleep - Fatigue and Attentiveness, i've decided to add them to my evening questionnaire, which takes no more than 1-1.5 minutes. These 8 questions should complement my morning sleep questionnaire.

Which factors may affect sleep?

After sleep quality and quantity assessment is established, we can more precisely measure how different factors affect sleep. Factors for testing:

I've slept 150 days with Dreem 2 and will analyze some of these factors.

Some of sleep disorders not too hard to diagnose. For example, obstructive sleep apnea (OSA) cause drops in SPO2 during night, which can be captured by pulse oximetry during night. For example Wellue SPO2 ring were accurate enough to catch my mother nightly SPO2 drops:

Using same ring on me shows clear picture with SPO2 never goes less than 95%.