
RemKeeper
Date: Now
Role: TBD
Range: Ideation to Prototype
Collaborators: ​TBD

BACKGROUND
​
Recently, I found myself increasingly grappling with postprandial somnolence. The bouts of sleepiness after meals grew so compelling that taking a short nap of about 30 minutes on a bean bag became my routine. Known as a Power Nap, this brief slumber remarkably recharged me, sharpening my focus for the remainder of the day's work. However, working in a shared office like WeWork, using a loud alarm was impractical. Instead, I relied on the haptic alarm feature of my Apple Watch. Waking up to a gentle vibration rather than a jarring sound proved to be an ideal solution for me, enhancing my napping experience without disturbing others.
After resigning from Charco to concentrate on my MBA studies and prepare for a new life in the United States, I joined the KAIST i-Corps startup team program. Here, I collaborated with individuals passionate about engineering and design to brainstorm and initiate a new project. Deeply immersed in the digital healthcare industry through my experiences with Keeper projects and Charco, I was eager to expand my expertise. I was particularly intrigued by the potential of leveraging health data in the era of generative AI to offer valuable services. My interest wasn't confined to posture correction or managing neurological conditions like Parkinson’s; I was also keen on exploring sleep quality.
Among the services I frequently used was an app called Sleep Cycle Alarm. Launched in 2009, I had been a consistent user ever since discovering it. The app works by tracking the user's sleep patterns and aiming to wake them up during the lightest phase of sleep. Inspired by Sleep Cycle Alarm's principle and the haptic technology utilized in both Keeper and Charco devices, I pondered over a device that could track users' sleep data and gently wake them up with a gradational vibration during their lightest sleep phase. After receiving positive feedback from my peers on this concept, I decided to develop a prototype.



DESIGN PROCESS - ENGINEERING
The engineering aspect of the RemKeeper project was centered on two core elements: real-time monitoring of a user's sleep stages and providing appropriate haptic feedback during the lightest phase of sleep within a designated time frame (e.g., +/- 30 minutes from the set alarm time).
For sleep stage detection, we opted to use the EMFIT sensor, produced by a Finnish company specializing in sleep trackers since 1993. Renowned for their extensive research, with over 100 scientific publications, EMFIT utilizes ballistocardiography—a technique for sensing the mechanical activity of the heart and breathing movement analysis—to gather sleep data. This method aligned perfectly with our vision for a non-invasive design.
Next, we focused on how to adjust the vibration features (strength, pattern, frequency, etc.) during each sleep stage to ensure the user awakens with minimal disturbance. Leveraging my experience in the Charco project, where we conducted clinical trials and evaluations to match the optimal CUE1 haptic technology with various symptom types of Parkinson’s patients, I planned to apply similar principles to user sleep testing for RemKeeper. Additionally, the EMFIT sensor measures more than just sleep stages; it also provides real-time data on heart rate variability (HRV), breathing rate, and more. We are currently devising a service that utilizes this comprehensive data to enhance users' sleep quality and developing the necessary algorithms to support this functionality.



DESIGN PROCESS - ENGINEERING & DESIGN FOCUS
​
In the development of RemKeeper, our engineering and design focus was on creating a non-invasive yet effective solution to provide haptic feedback closely to the user's body. We envisioned a slim, elongated rectangular device to be placed under the mattress. Drawing from user experience with similar devices like Withings Sleep, which has a relatively thick sensor and PCB causing discomfort when turning in bed, we aimed for an improved design. Inspired by the Beddit product, now acquired by Apple, we considered positioning the main PCB in the adapter part of the device, minimizing the thickness and potential discomfort of the in-bed components.
​
While the EMFIT sensor excels in measuring specific sleep stages, it doesn't seem optimal for detecting the kind of user movement or actions, such as vibrations, that we are planning to incorporate. Its precision in tracking sleep-related movements appears limited, seemingly only capable of basic determinations like bed exits. Consequently, we are considering the inclusion of a 6-axis accelerometer, which could offer more accurate detection and analysis of the user's motions during sleep.
​
In a significant design shift, we decided to eliminate the power button, inspired by the lessons learned from the Keeper project. By designing the device to automatically switch to a power-saving mode when no user is detected and reactivate upon return, we could eliminate the need for manual power control and further miniaturize the device. Nevertheless, we recognized the necessity of an LED indicator to convey the device's status, such as power and pairing, ensuring user awareness and interaction with the device's operational state.



DESIGN PROCESS - APP & SLEEP DATA-BASED SERVICE
At the heart of this project lies the application services and content based on user data, more so than the device itself. The RemKeeper, integrating hardware and software, offers three main functionalities:
Rem Tracking: Beyond collecting user data from sensors, such as Sleep Stage and Heart Rate Variation, RemKeeper also utilizes the smartphone's microphone to gather comprehensive sleep data. This includes breathing disturbances during various sleep stages, providing an overall view of the user's sleep quality.
​
Rem Sleep: The device aims to reduce the user's sleep latency period by emitting ultra-low-frequency vibrations, known for inducing deep sleep and relaxation, particularly in the delta (1 to 4 Hz) range. Furthermore, it can play sleep-aiding sounds, such as pink noise, through the smartphone's speaker, enhancing the sleep environment.
Rem Alarm: RemKeeper estimates sleep stages in real-time and gently awakens the user during the lightest phase of sleep (known as NREM stages) with optimized vibrations. Concurrently, it can activate awakening sounds (ideally within 100 – 120 beats per minute) through the smartphone's speaker, facilitating a gentle and pleasant waking experience.
​
The RemKeeper project, with these three functionalities, aims to revolutionize the entire sleep experience for the user, from pre-sleep preparation, through the sleep phase, to post-sleep awakening. It strives to dramatically improve and empower users to self-manage their overall sleep health.