Smartwatches and rings that claim to measure sleep estimate the phases based on heart rate and movement, but this estimation is not very accurate. SleepOne proposes measuring brain electrical activity using an ear-worn device that functions as an electrode within the ear. In this interview with Toque de Ingenio, David Ferri explains what signal a device of this type needs, how the electronics have been designed to function throughout the night, and what it means to validate and fund a health product from Spain.
Guest: David Ferri, who was introduced in the episode as a biomedical engineer leading SleepOne. Interview published: January 20, 2026. Duration: 1 h 1 min.
In this episode:
- Why heart rate and movement alone are not enough to recognize the stages of sleep.
- How a conductive silicone earbud can record an electroencephalogram (EEG) without adhesive electrodes.
- What electronic, battery, and data-related decisions enable its use every night.
When the conversation was recorded, the device was not yet available for purchase. David had been working on the project for over two years, and the algorithm was awaiting clinical validation.
Why does a clock not measure sleep?
David Ferri begins by differentiating between two types of product. A wellness device can show how you slept, without that being considered a medical diagnosis, and therefore can be sold even if the measurement isn’t reliable. As he explains, scientific studies demonstrate very low accuracy in these sleep patterns, but users are guided by marketing.
The underlying problem is the signal. Watches, bracelets, and rings measure the same thing: heart rate and movement, and using these two variables allows us to approximate sleep stages. David acknowledges that looking for this relationship was a reasonable idea, but he asserts that it has already been proven that it doesn’t work.
Sleep takes place in the brain and can therefore only be measured by observing its electrical activity. This is the signal doctors analyse and the scientific community considers the reference. According to David, the obstacle was the lack of an affordable, comfortable way to record it every day outside hospital.
“Sleep happens in the brain; in other words, sleep is simply the brain doing things.”
David Ferri 10:16.
Measuring sleep in the ear: conductive silicone instead of electrodes
The clinical alternative is a sleep study: a helmet with electrodes, cables, and sensors that a doctor analyzes manually. David confirms that this is currently the only way to determine how someone sleeps, and it can be counterproductive for those who already struggle to fall asleep.
SleepOne doesn’t aim to replace all of that testing. David clarifies that apneas are pauses in breathing that aren’t detected by the brain, and for these, there are more reliable devices. His estimate is that between 60 and 70 percent of the problems studied using the electrode helmet could be covered by simply measuring brain activity.
The device is a headphone that does not play music. Three zones made of a special silicone with metallic particles make it conductive, effectively functioning as an electrode. When placed in the ear, the voltage reaching the device is the one produced by the brain in that area.
Choosing the ear over the forehead is based on the principle of adhesion. In other areas of the head, a disposable adhesive electrode is needed – a consumable that the user eventually abandons. However, with the ear, it’s possible to use something that can be easily removed and replaced, which David considers to be the best way to establish a daily habit. Existing straps are accurate, but they cost around 500 euros and are uncomfortable to use while sleeping.
The architectural structure that the algorithm must recognize.
David clarifies that he is not a neurophysiologist or a doctor, and that his explanations are those of an engineer working with specialists. Using this caution, he describes the architecture of sleep: initially, one enters a light sleep, followed by deep sleep, and REM sleep appears primarily towards the end of the night. It is normal for these cycles to be repeated two or three times without waking up.
Each phase leaves a distinct mark on the signal. Upon waking, the electroencephalogram is chaotic, with low amplitude and high frequency. During deep sleep, the brain slows down, and the signal increases in amplitude and decreases in frequency. REM sleep resembles wakefulness to such an extent that it’s almost impossible to distinguish them when looking only at the waveform. According to David, deep sleep is associated with tissue repair, and REM sleep with memory consolidation; he himself requests that a specialist correct any figures that are given.
For development, the key is how that signal is transformed into information. An algorithm classifies each 30-second segment as being awake, lightly asleep, deep sleep, or REM sleep, and the result is the hypnogram – which the bracelets already display – but based on actual data. David explains that this algorithm will be validated in a scientific study with several hospitals, comparing it to the diagnoses of clinical experts.
Electronics for a whole night without mobile phone access.
When asked about the technical difficulty of measuring brainwaves, David responded that it’s not science fiction. He explained that the system uses three silicone electrodes: one serves as a reference point, and the other two are active. A chip calculates the difference in voltage between each active electrode and the reference, and then the difference between those two results – representing the surface voltage on the skin.
Only one earbud is registering. The other is a passive silicone plug for those who already use earplugs. An integrated accelerometer detects movement and, combined with the change in the brain signal, allows the user to identify when they get up.
David emphasizes that the device does not transmit data in real-time via Bluetooth for two reasons: to conserve battery during a long night, and to avoid wearing a transmitter on the head. The device records the night’s data and downloads it to the mobile phone while it’s charging in its case. The aim is to ensure at least 12 hours of autonomy, as elite athletes require longer sleep for recovery.
These decisions illustrate how the Electronic engineering of a medical wearable is driven by its intended use: power consumption, memory, sensors and charging all depend on how the device will be used.
Elite athletes: moving from feelings to objective data.
David’s initial focus is on sports clubs. His aim is to develop patterns that allow for predicting injuries based on reliable daily data – something that doesn’t currently exist. Top-level medical teams, he notes, currently work with forms focusing on how the player has slept – purely subjective data.
According to his account, the clubs he has spoken to understand the proposal fully. The obstacle lies in the budget: there are seasons when this ambition isn’t included within the metrics the club decides to invest in. For the end-user, the model discussed during the interview involved a monthly subscription of around €30, without the need to purchase the device beforehand.
David accepts that, in the short term, the value for the user lies in adapting routines. However, he sees the greater value in the long term, with data series that could potentially relate to age-related disorders. This is an expectation presented by the guest, not a proven result.
Financing medical hardware from Spain.
According to David, the biggest challenge is securing the funds needed to develop the device. An engineer without savings is financing the project using their savings and a full-time salary, dedicating all their free time to it.
“We rely on the finished device to be able to invoice”.
David Ferri 54:59.
That phrase encapsulates the challenge faced by a physical product when competing against a business that generates revenue from day one. In Spain, it’s difficult to find investors willing to invest in the early stages of hardware. The company has recently had to seek capital outside the country, largely through LinkedIn, while awaiting the results of clinical studies.
Regarding his biggest mistake, he responds that he would have transitioned to full-time work from the outset. In his opinion, what currently takes three or four years of juggling two jobs could be resolved in just one, and investors view a full commitment more favorably.
What can another development team learn?
The SleepOne case offers four key lessons applicable to other wearables and health devices:
- Measure the correct variable. An indirect signal can yield a marketable product, but it may not actually answer the question the user is asking.
- Design for adhesion. The measurement area and the “add and remove” format were chosen to ensure that daily use doesn’t rely on disposable items.
- Reduce unnecessary functions to gain an advantage. Disabling Bluetooth in real-time simplifies the electronics and protects the battery.
- Treating validation as an integral part of the product. The clinical study conditions the trust of investors and clients just as much as the design.
At the time of the interview, the SleepOne device was not yet available for sale, and the algorithm was awaiting validation through several hospitals.
In i-mas, we accompany product development and the prototyping of devices that need to capture small signals, operate on battery power, and pass a validation process. If you’re working on wearable technology or a portable medical device Tell us about your project..
To continue reading: How does one develop a medical device with its own electronics?.
Source of article: Interview with Edgar Guerrero on Toque de Ingenio with David Ferri. The excerpts cited link to the minute of the conversation. The recommendations for other projects are an editorial summary from i-mas.
