Project overview
A solution providing accurate measurement of user’s sleep metrics, including total sleep duration, boundaries and duration of sleep phases, such as REM, light sleep and deep sleep; sleep efficiency and sleep interruptions.
Customer profile
In-house proprietary project
Challenges 
- Need to design a sleep data analysis technology with high aссuraсy of sleep tracking based on limited sources of data
- Need for compact design of the wearable device
- Limited computing capabilities of the microcontroller user in the wearable device.
Technical highlights 
- Developed data acquisition techniques for a physical sensor
- Developed data filtration and pre-processing algorithm
- Data analysis and model definition enabled with the help of BaseGroupLab Deductor Studio software
- Smart alarm – identification of an optimal wake-up moment
- Developed the model for sleep phases detection
- Developed an algorithm for sleep tracking, using correlation analysis and cluster analysis.
Business value
- According to lab test results, the average accuracy of detecting REM sleep phase boundaries is 73% (market median is 70%).
- Implemented advanced data visualization techniques as a key differentiating function
- Accuracy was verified based on the indicators of professional medical equipment, such as Nihon Kohden NEUROFAX EEG-1100K electroencelograph and TrackIt (TM) Sleep Walker(TM) polysomnograph.