Introduction

Here is my experience in using some wearables, apps and my own tools.
Statistical analysis for most of statements below already done and will be added soon.

Oura ring 2.0/3.0

I'm wearing Oura ring 2.0 every day and night. I've analyzed data for about 1 year and pretty confindent in these statements:

• Sleep stages and total sleep time (TST) seems to be incorrect, not reaching required accuracy
• Awake time (WASO) and sleep onset latency (SOL) also innacurate
• Only one sleep metric seems to be accurate - Time in bed (TIB).
• Restlessness which represent movement during sleep have a poor correlation with another wearables
• HR and HRV for a whole night seems to be pretty accurate and used in couple of studies [1] [2] [3]
• Temperature variation is accurate enough to detect sickness
• Not sure about breathing rate, but it have strong correlations with other devices and reflect periods of illness
• Steps and activity - overcounted, but reliable and correlate pretty well with another wearables. Usually i do multiply steps by 0.8 to get something not far from reality
• Simple and easy Web API access

Oura have acceptable accurate to answer these questions: How much time i spent in bed? Do i have any signs of sickness (temperature, hr, hrv, breathing rate)? how did my training impact nightly restoration (hr, hrv)? How active i was that day (steps)?

UPDATE: i've finally got Oura ring 3.0 and i see that daily HR in some degree is ok. SPO2 has 1-day resolution which is unuseful.

Fitbit Charge 4/5

I'm wearing Fitbit Charge 4 every day and night. I've analyzed data for about 4 month and pretty confindent in these statements:

• Sleep stages and total sleep time (TST) seems to be incorrect, not reaching required accuracy
• Time in bed (TIB) looks good
• HR and HRV looks noisy, but averaging works fine. HRV have moderate correlation with Oura ring
• SPO2 Variations and average nightly inaccurate
• Steps and activity seems to be better then Oura
• Skin temperature extremely noisy. Its hard to see circadian pattern.
• Web API access is complicated, but possible
• Request data provides a lot of raw data in CSV which isnt presented in app.

Questions: How much time i spent in bed? Do i have any signs of sickness (temperature, hr & daily hr patterns, hrv, breathing rate)? how did my training impact nightly restoration (hr, hrv)? How active i was that day (steps, daily hr)?

There is 2 main reasons of using Fitbit Charge 4: complement oura data and get daily HR in 1 minute resolution.

UPDATE: right now i'm using Fitbit Charge 5. It's seems to be more comfortable and lighter, everything else is the same.

Polar H10

Polar H10 is a most scientifically validated device in my collection.
I use everyday for tracking of my trainings, measuring Standing HRV at morning and doing submaximal tests for HRRecovery and VO2Max proxy tests

• Perfect HR and HRV data according to wide scientific research
• Web API access is complicated, but possible
• Easy export to TCX & GPX

Questions: How much did i train (time, hr, rpe, training load)? Whats my fitness status (hr during standardized tasks for vo2max, hr-ex and hr-recovery)? How my ANS deals with current amount of physical load and illness (morning standing hrv)?

Dreem 2 Headband

Using my Dreem 2 for ~40% of time (145 nights over last year).

• Most accurate sleep metrics assessment available for consumer
• Acceptable comfort, after few days of use
• Some parts easy to brake and need DIY rework
• Easy export summary to CSV, including 30-sec resolution hypnogram
• No access to raw EEG data

Since Dreem left consumer market there is no easy alternative. Hypnodyne ZMax is huge in pricing and works only with Windows.

Questions: What's my sleep quality and quantity? How far i'm from sleep disorders? Whats my sleep trends?

Kinsa Smart Thermometer

I'm using Kinsa QuickCare every day to asess morning mouth temperature.

• Mouth temperature seems to be pretty accurate with ±0.1C variance
  
• Strong correlations with Oura nightly temperature deviation
• Very fast (about 10 secs), but need to be correctly located in mouth with closed lips
• Showing circadian rhytms if used for a long time, but not so sensitive as iButton
• Easy export to txt, possible conversion to CSV with regex some values
  

Questions: Am i sick? How my temperature changes through year?

Wellue SPO2 Ring

After i've used Wellue SPO2 ring every night for about 45 nights i didnt found any issues and see my data pretty stable. After realizing that i've switched to occasionally use when its needed: chances of being sick soon, already sick and on random days. Oura ring 3.0 may totally replace this ring if it adds 1-minute resolution for SPO2

• Seems to be most accurate SPO2 device with high resolution (4 sec), acceptable
• HR data have excellent correlation with Oura ring
• I never used or compared movement data, but its present. Not sure about accuracy
• No API, but easy export to CSV

Questions: Am i sick with respiratory diseases? Do i have sleep apnoe? How my SPO2 drops during flights?

FreeStyle Libre

I'm using this FreeStyle Libre V1 sensor for 2 weeks every 1-2 months to assess my glucose status.

• I'm using Glimp as main app and export data into CSV
• Sometimes it hurts a little when you set it up for a 1-2 days
• Sometimes its having some bias of real glucose
• Trend seems to be pretty accurate

Usually, after 2 weeks, i export my data into CSV and upload for glycotype assessment here. My target is being in Low group
Questions: How far i from metabolic disorders?

Withings Sleep v1 & v2

• HR data seems to be ok, but sometimes may be noisy
• Sleep staging is far from EEG, but not worse than most wrist trackers.
• Time in bed / sleep start time seems to be accurate.
• Located under mattress and tracks passively without any contact.

Withings BPM Core

I'm using it to control my blood pressure every morning. ECG & BPM data available through API and seems to be accurate.

• Accurate blood pressure
• Stethoscope doesnt seems to be useful and fails most of time because of ambient sounds disturb signal.
• ECG feature seems to be ok.

Questions: since my BP on a lower end, i want to be sure my BP doesnt go too low.

Withings BPM Connect

Alternative to BPM Core during travels. Measures only blood pressure and HR. Data available through API.

App for tracking events

I've wrote simple telegram bot which can

• Ask a questions showing some buttons for different answers and stores anwers
• Can accept messages with text and numeric values at the end and stores them too
• On daily and weekly basis bot sends me morning & evening questionnaires about my mood, sleep quality etc with LikeRT buttons. It costs me 2 min per day and 7 min per week.
• I send bot my symptoms and manual measurements. <1 min per day.

Doing that for about 1 year i got 40 000 of different events available for deep analysis. Summary all this manual tracking takes about 25 minutes per week, but provides inside in subjective sleep, mood, fatigue, stress and manual anthropometric metrics like waist circumference.

I'm using Ornament App for biomarkers tracking

• Just import PDF from lab into app
• It recognizes all results and they instantly available in app with graphs
• Results can be export into PDF which can converted to table format with MS Word and then copied into google sheets and easily formatted into table allowing CSV export on next step

I've written simple service which process CSV files from all my wearables and imports it into MySQL database. Manual export CSV through apps takes just a few mins per week. If device have an API - its imported into database directly without handwork.

I'm rarely doing Psychomotor Vigilance Task for tracking my attention, potential sleep defficiency, sleep inertia and excess fatigue. App is Vigilance Buddy (2m duration, interstimulus 0.3-3s and 1s feedback, ~45 taps per test). My usual average response is 300ms and standard deviation (SD) is around 25ms and i want to detect at least medium effect sizes with significance of 0.95 and power of 0.8. Power analysis result is 40 taps:

Since i dont want to spent more than 2 minutes and less than 2 minutes maybe not enough i choose 2 minutes. Also i want to have a few additional taps (sometimes i have 1-2-3 false starts), lets assume 45 taps in 2 minutes which results in 120/45=2.6 average delay between responces and leads to 0.3-3s / 1s configuration (0.3 + (3 - 0.3)/2 + 1) = 2.65.

Stopped doing

Stool Log - Bowel Movement app used to track bowel movements with bristol stool classification. Easy export into CSV. Reason to stop: 

I'm occasionally doing Dual N-Back for tracking and improve my working memory, but not yet analyzed scientific background for it.

I'm in using Spaced Repetition technique with Anki software on a daily basis to learn in more productive way and assess long term memory performance.

All this data will be complemented with wearables data (sleep, hr, hrv, weight, bmi, temperature, blood pressure, activity, training load etc) and deeply analyzed to answer my questions and replicate some studies and check if evidence work on me.

Example of questions: Whats time is optimal for sleep? How not to overtrain? How physically active day impacts my sleep compared to sedentary day?

iButtons

Not using right now. I'm occasionally used ibuttons for tracking illness and assess circadian patterns.

• Pretty accurate for circadian rhytms distinction
• Useful for checking trends during illness
• Not convenient for regular daily use
  Export into CSV

Questions: How strong in my circadian rhytm? How my temperature changes during illness?

Reason to stop: Looking for easier ways to measure circadian rhytms. Kinsa thermometer is enough for ilness tracking.

Simple Arithmetics Test

Not using right now. I'm occasionally doing Simple Arithmetics Test mentioned by Seth Roberts which takes 2 minutes per day to assess attentiveness, reaction and maybe some basic cognitive / memory skills. There is conflict between goal for faster response (leads to increased error rate) and goal for lowering error rate (leads to slower response speed) which may lead to inconsistency in results. PVT doesnt have this problem.

Reason for stopping:  haven't found how to utilize results for decision making.

Atmotube PRO

Not using right now. I've used Atmotube Pro to track enviroment air quality. I'm not wearing it with myself everywhere, most of time it stays at home.

• Battery life for about week
• Not big and easy portable
• Have some validation [4]
• Easy export to CSV

Questions: How is air quality affects my sleep? How ambient pressure affects my blood pressure?

Reason to stop:  haven't found how to utilize results for decision making.

MetaMotionR

Not using right now.

• no HR, but have some interesting sensors (for example ambient light)
• very buggy software
• have export to CSV

Reason to stop:  haven't found how to utilize ambient light for decision making. There are better alternatives

Polar Vantage V2

Not using right now. I've bought it mainly for orthostatic HRV test and trainings when i dont have access to Polar H10. After checking some science i've realized that i dont need supine HRV which takes 2 of 4 minutes in Polar orthostatic test. I've switched to using H10 for 1 min stabilization period and 1 min measuring window during standing position with Kubious App.

• HR was less accurate compared to Fitbit and Polar H10.
• Nightly HRV looks noisy and there is moderate correlation with Oura
• Sleep data isnt accurate for sleep staging
• Too big and not convinient
• Web API access is complicated, but possible

Reason to stop:  The only unique feature was the orthostatic test which i've replaced with standing HRV test.

Garmin Vivosmart 4S

Not using right now

• HR was less accurate compared to Fitbit and Polar H10.
• SPO2 inaccurate (seems true for all wristbands).
• Stress metrics doesnt make sense because based on inaccurate HR which results in inaccurate HRV
• No free API, only for huge businesses

Reason to stop:  Fitbit Charge 4/5 is more accurate and comfortable.

Withings GO

Not using right now

• Used in past few years
• Very accurate steps when you walk (validated with counting)
• No extra steps when you are not taking steps
• Using as a clip with hands free
• Web API access is complicated, but possible
  Data export in CSV is possible

Reason to stop:  stopped using after i've realized i want 24/7 HR & some HRV data.

P.S I dont have any affiliations with any these manufacturers