Key Concepts

🙋‍♀️ User

User is defined as a specific person you’d like to record sleep. You can use the SDK or API to generate a user uuid that you want to record your sleep. This allows you to manage sleep sessions on a per-user basis.

🛌 Sleep Session

A sleep session is a concept that corresponds to one’s sleep(lying in bed until waking up). You can create a sleep session for a specific user, upload the data you want to analyze, and end the session.

The session has a total of three states: OPEN, CLOSED, and COMPLETE.

  • OPEN: Session has been created and audio data can be uploaded for analysis
  • CLOSED: No more audio can be uploaded at the end of the session, waiting for the analysis to complete
  • COMPLETE: All analysis is completed

📇 Mel Spectrogram

Audio data measured on a smartphone or smart device is converted to the form of a Mel spectrogram through the SDK's preprocessing algorithm and sent to the server.

Mel Spectrogram

Mel Spectrogram

📊 Data Metrics

If you use the real-time analysis feature, you can check the sleep stages, breath stages, and snoring stages analyzed to date in real-time for any Open session. When sleep ends and all analyses are complete, the data available from the analysis results are as follows.

[ Explore Data Offerings by Plan ↗️ ]

1. Basic Sleep Information

Data collection related to the start and end of sleep measurement, as well as sleep onset and wake-up times.

  • Sleep latency (sleep_latency) and Wakeup latency (wakeup_latency) can be used as data to infer the user's habits (e.g., difficulty falling asleep, lying in bed for a long time after waking up).

  • Total Measurement Time (time_in_bed) and Sleep Period Time (time_in_sleep_period) serve as the standard for different data metrics, so be careful to distinguish between the two data sets.

No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1start_timeTimeExact time at which the tracking started.2023-11-22T00:00:10+09:00 [YYYY-MM-DDThh:mm:ss±hh:mm] (=2023-11-22 00:00:10)
2end_timeTimeExact time at which the tracking ended.2023-11-22T06:30:38+09:00 [YYYY-MM-DDThh:mm:ss±hh:mm] (=2023-11-22 06:30:38)
3time_in_bedDurationThe time duration between the start and the end of tracking.23428 [s] (=6h 30m 28s)
4sleep_timeTimeExact time at which ‘Sleep’ was first detected since the tracking started.2023-11-22T00:53:10+09:00 [YYYY-MM-DDThh:mm:ss±hh:mm] (=2023-11-22 00:53:10)
5wake_timeTimeExact time at which the last 'Wake' period was detected during tracking. No sleep will be detected after wake_time.2023-11-22T06:30:08+09:00 [YYYY-MM-DDThh:mm:ss±hh:mm] (= 2023-11-22 06:30:08)
6time_in_sleep_periodDurationThe time duration between the onset of ‘Sleep’ and the start of the last ‘Wake’ during tracking.20190 [s] (= 5h 36m 30s)
7sleep_latencyDurationThe time duration between the start of tracking and the detection of first 'Sleep'.3180 [s] (=53m 0s)
8wakeup_latencyDurationThe time duration between the start of the last 'Wake' and the end of tracking.30 [s] (= 0m 30s)
9sleep_indexScoreThe metric that comprehensively represents sleep quality, defined by learning from the distribution of sleep data.97 (range: 50 ~ 100)

1-1. Sleep Score

The sleep_index is defined by three indicators: total sleep time (time_in_sleep), which measures the quantity of sleep; sleep efficiency (sleep_efficiency), which measures the quality of sleep; and the number of awakenings (waso_count). The sleep_index ranges from 50 to 100 points and is designed to have the highest correlation with users’ subjective sleep satisfaction, based on medical definitions of sleep quality and sleep big data.

Awakenings during sleep are categorized into long awakenings that a person can perceive and short awakenings lasting 30 seconds to a few minutes, which are difficult to notice. Short awakenings that are distributed throughout sleep are reflected in the score through waso_count, while long awakenings that a person can perceive are reflected through sleep_efficiency.


2. Sleep Efficiency

Data related to Sleep Efficiency (sleep_efficiency), one of the key indicators representing the quality of sleep.

  • By figuring out the Total Sleep Time (time_in_sleep) during the Total Measurement Time (time_in_bed), you can calculate Sleep Efficiency (sleep_efficiency).
  • While there is no one-size-fits-all answer for sleep, having a high Sleep Efficiency suggests securing an optimal amount of sleep in relation to the time spent in bed.
  • While the Sleep Stage Ratio (sleep_ratio) data is based on the Sleep Period Time (time_in_sleep_period), please note that Sleep Efficiency (sleep_efficiency) data is based on the Total Measurement Time (time_in_bed)
No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1time_in_sleepDurationThe actual time spent sleeping. It is the time duration calculated after subtracting all ‘Wake’ time from time_in_bed.19380 [s] (= 5h 23m 0s)
2sleep_efficiencyRatioThe ratio of time_in_sleep to time_in_bed0.83 [x.xx] (=83%)

3. Sleep Stage

Data collection related to the four sleep stages: Wake, Light Sleep, Deep Sleep, and REM Sleep.

  • During the Total Measurement Time (time_in_bed), the user's sleep is recorded every 30 seconds as one of the four stages. You can check this collection of sleep stages through the Sleep Stage List (sleep_stages)
  • The frequency and duration of awakenings during sleep can be crucial indicators affecting the subjective satisfaction of sleep, making them important metrics during result analysis. Relevant data, such as Wake After Sleep Onset Count (waso_count), Time in Wakeful Stage (time_in_wake), and Longest Wake After Sleep Onset (longest_waso), can be utilized for this purpose

No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1sleep_stagesNumber ListA list of numbers, indicating sleep stages during time_in_bed
(0: Wake / 1: Light / 2: Deep / 3: REM)
[0,0,1,……,2,2,3,3,1,1,0,0,0] [list] (= wake, wake, light, .....)
2waso_countCountThe number of ‘Wake’ detected during time_in_sleep_period19 [#] (= 19 times)
3longest_wasoDurationThe longest time duration of ‘Wake’ during time_in_sleep_period90 [s] (=1m 30s)
4time_in_wakeDurationThe sum of all ‘Wake’ seconds detected during time_in_sleep_period810 [s] (=13m 30s)
5time_in_lightDurationThe sum of all ‘Light’ seconds detected during time_in_sleep_period8610 [s] (=2h 23m 30s)
6time_in_deepDurationThe sum of all ‘Deep’ seconds detected during time_in_sleep_period5700 [s] (=1h 35m 0s)
7time_in_remDurationThe sum of all ‘REM’ seconds detected during time_in_sleep_period5070 [s] (=1h 24m 30s)

4. Sleep Stage Ratio

Data collection related to the ratio of awakenings and each sleep stage during the Sleep Period Time (time_in_sleep_period)

  • The Sleep Stage Ratio (sleep_ratio) data represents the proportion of time recorded in Light, Deep, and REM stages during the Sleep Period Time (time_in_sleep_period), excluding wakefulness. To compare the ratio of awakening to sleep in the user's sleep period time, you can utilize both Sleep Stage Ratio (sleep_ratio) and Wake Ratio (wake_ratio).
  • To gain an overview of the user's overall sleep structure, you can compare and utilize the following four data: Wake Ratio (wake_ratio), Light Sleep Ratio (light_ratio), Deep Sleep Ratio (deep_ratio), and REM Sleep Ratio (rem_ratio).
No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1sleep_ratioRatioRatio of time_in_sleep to time_in_sleep_period0.96 [x.xx] (=96%)
2wake_ratioRatioRatio of time_in_wake to time_in_sleep_period0.04 [x.xx] (=4%)
3light_ratioRatioRatio of time_in_light to time_in_sleep_period0.43 [x.xx] (=43%)
4deep_ratioRatioRatio of time_in_deep to time_in_sleep_period0.28 [x.xx] (=28%)
5rem_ratioRatioRatio of time_in_rem to time_in_sleep_period0.25 [x.xx] (=25%)

5. Sleep Stage-specific Latency

Data collection related to the time it takes for each sleep stage (Light Sleep, Deep Sleep, REM Sleep) to first appear from the onset of sleep.

  • Generally, as the Sleep Latency (sleep_latency) increases, the length of the REM Sleep Latency (rem_latency) also increases.
  • Since the first detected sleep after the user initiates sleep tracking is always Light Sleep, please note that, barring any errors, the value for Light Sleep Latency (light_latency) data is always 0.
No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1light_latencyDurationThe time duration between sleep_time to the first detection of ‘Light’
(Normally, '0' as light sleep is the first sleep stage to be detected)
0 [s] (=0m 0s)
2deep_latencyDurationThe time duration between sleep_time to the first detection of ‘Deep’540 [s] (=9m 0s)
3rem_latencyDurationThe time duration between sleep_time to the first detection of ‘REM’4710 [s] (=1h 18m 30s)

6. Sleep Period

Data collection related to the sleep cycles that periodically occur within the Sleep Period Time (time_in_sleep_period).

  • Typical sleep involves a pattern of alternating between Non-REM (NREM) and REM sleep. A bundle of repeating NREM and REM sleep is referred to as a sleep cycle, and the start and end times of each sleep cycle can be confirmed through the Sleep Cycle Transition Time List (sleep_cycle_time) data.
  • A healthy adult typically experiences 4-6 cycles of sleep per night, each lasting 70-110 minutes, when obtaining around 8 hours of sleep. To comprehend the user's sleep cycle structure, you can utilize the Sleep Cycle Count (sleep_cycle_count) and Average Sleep Cycle Duration (sleep_cycle) data.
No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1sleep_cycleDurationThe average time duration of sleep cycles. Each sleep cycle is consisted of both 'REM' and 'Non-REM' sleep.6520 [s] (=1h 38m 40s)
2sleep_cycle_countCountThe number of sleep cycles detected during the time_in_sleep_period3 [#] (=3 times)
3sleep_cycle_timeTime ListA list of the exact start and end times of each sleep cycle[2023-11-22T00:53:10+09:00,2023-11-22T02:26:10+09:00,2023-11-22T04:54:10+09:00,2023-11-22T06:19:10+09:00] [list] (= 2023-11-22 00:53:10, 2023-11-22 02:26:10, 2023-11-22 04:54:10, 2023-11-22 06:19:10)

7. Unstable Breath

Data collection related to understanding the user's respiratory stability and instability during sleep.

  • Similar to the Sleep Stage List (sleep_stages) data, the respiratory stages are recorded as 0 (Stable) and 1 (Unstable) during the Total Measurement Time (time_in_bed). This collection of respiratory stages can be confirmed in the Breath Stage List (breath_stages) data after the measurement is concluded.
  • The user's irregular breathing is converted into Unstable breathing index (breathing_index) data, and the severity of the Unstable breathing pattern (breathing_pattern) is determined by its value.

No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1breath_stagesNumber ListA list of breath stages during the time_in_bed
(0: Stable / 1: Unstable)
[0,0,0,1,1,……,1,0,0,0,0] [list] (= stable, stable, stable, unstable,.unstable.......)
2unstable_breath_countCountThe number of times unstable breath is detected during the time_in_sleep_period11 [#] (=11 times)
3time_in_unstable_breathDurationThe time duration, of which unstable breath is detected during time_in_sleep_period2400 [s] (=40m 0s)
4time_in_stable_breathDurationThe time duration, of which stable breath is detected during time_in_sleep_period17790 [s] (=4h 56m 30s)
5unstable_breath_ratioRatioThe ratio of time_in_unstable_breath to time_in_sleep_period0.12 [x.xx] (=12%)
6stable_breath_ratioRatioThe of time_in_stable_breath to time_in_sleep_period0.88 [x.xx] (=88%)
7breathing_indexNumber IndexThe numbers indicating the level of breathing instability9.33
8breathing_patternNumber IndexA list of names indicating the level of breathing instability:
'Stable', 'Slightly Unstable', 'Unstable', 'Severely Unstable’
MILDLY_UNSTABLE_BREATH (= Mildly Unstable)

8. Snoring

No.Data NameTypeDefinitionExample [Units] (= Actual Value)
1snoring_stagesNumber ListA list of snoring stages during the time_in_bed
(0: no snoring / 1: snoring)
[0,0,0,1,1,……,1,0,0,0,0] [list] (= no snoring, no snoring, no snoring, snoring, snoring .......)
2time_in_snoringDurationThe time duration, of which snoring is detected during time_in_sleep_period23400 [s] (=6h 30m 00s)
3time_in_no_snoringDurationThe time duration, of which no snoring is detected during time_in_sleep_period5400 [s] (=90m 0s)
4snoring_ratioRatioThe ratio of time_in_snoring to time_in_sleep_period0.72 [x.xx] (=72%)
5no_snoring_ratioRatioThe ratio of time_in_no_snoring to time_in_sleep_period0.28 [x.xx] (=28%)
6snoring_countCountThe number of times snoring is detected during the time_in_sleep_period20 [#] (=20 times)