A Word (Or Several) On Nicola Bulley's Fitbit
Originally posted on Twitter on June 29
What follows is far too many words on Nicola Bulley's Fitbit. I'm going to slightly out myself as a former director at a major technology company that makes fitness wearables. My role saw me entrenched in product development, design, product testing, marketing, user manuals, customer service, the list goes on. This is still the field I work in today. It's why I've been so fixated on the Fitbit throughout this case.
The narrative we were given for this Fitbit just isn't possible. It's either incomplete, as provided at the inquest, based on assumptions that haven't been thoroughly tested, or both. First, what is possible from the narrative we were given:
1. A Fitbit could pick up a heart rate from water moving between the device and wrist, or from other movements of a body. The LED technology that measures heart rate by blood flow in the capillaries in the wrist is not a perfect technology. In competitive fitness circles it’s widely accepted that wrist heart rate monitors (HRMs) are not as accurate as chest HRMs, which use different technology and multiple nodes on the chest to get a reading. Anecdotally, people writing on Fitbit's public forums have reported getting heart rate readings for Fitbit-wearing loved ones after they had been officially declared passed. Some have stated that heart rate was recorded when their loved ones' bodies were moved. One theory about this, unacknowledged by the company, is that the devices approximate a reading of heart rate based on historical averages even when a person has passed, when some kind of movement of the body occurs, and even in some cases when a device is not on a (living) person’s wrist.
2. That movement in water could be recorded as steps. Fitbits, like most other, newer wearables, use a three-axis accelerometer that is designed to provide the most accurate measure of actual steps. But it’s not perfect. If a body makes a movement that feels/looks like a step, Fitbit might record it as one, even if no (or minimal) distance was covered.
What I don't think is possible from the narrative we were given:
1. That her device, a Fitbit Versa 4, held a charge for 8 days submerged in ~3C water, and that the device was recoverable on Feb 19. The device is rated to a water depth of 50 meters, but its battery life (listed as "6+ days"), and arguably the device itself, would be compromised by being in cold water for this long. It is not designed for such use, nor are most wearables, except top-of-the-line models from Garmin, its direct competitors (e.g. Apple), and watches specifically designed for diving. Some sources, as Nick Bennett on Twitter found), give the Versa 4 an IP rating of 67. IP measures resistance to dust and water, and a rating of 67 allows for immersion in up to 1 meter for up to 30 minutes. But Fitbit told me that their devices don't have IP ratings at all (see image 1 below). 50 m is equivalent to 5 ATM, ATM being another rating that wearable and watch manufacturers use. This means it is water-resistant, but not waterproof, to a depth of 50 meters. I know from experience that IP ratings are costly for companies to obtain, but many premium fitness watch/wearable brands prioritize them.
The water in the Wyre was apparently 3 degrees Celsius on Jan 27 and somewhere in that range for the 23 days thereafter. The Fitbit Versa 4 uses a lithium polymer battery. Lithium polymer is the preferred technology for batteries used in smaller devices (vs lithium ion) due to its lighter weight, lower cost, and other advantages. But it performs worse in water and in extreme temperatures than lithium ion, which itself sees reduced battery life in cold and hot temperatures. The choice of LiPo *may* be a reason why Fitbit eschews IP ratings. LiPo just isn't that great for extreme conditions, or over the long term.
Wearable manufacturers are hard-pressed to get anything over 5-6 days in battery life—let alone 8—in normal operating conditions in devices that are actively recording, or attempting to record, heart rate. Based on everything I know about this technology, what they claimed at the inquest is remarkable, and that's putting it kindly.
2. That they obtained nine days of unsynced data when they retrieved the Fitbit on Feb 19. The inquest revealed details about Nicola’s walk, heart rate and last movements up until 9:30am on Jan 27, plus HR data recorded after 9:30am and until Feb 4. In total that is nine days’ worth of data. If Emma White’s comments to the media on Feb 4 (https://thesun.co.uk/news/21274369/nicola-bulley-cops-examine-fitbit-data/) are to be believed, the device was last synced to Nicola’s phone on Tuesday, Jan 24. The device does not sync to the cloud automatically. It requires a Bluetooth-linked phone to sync to the app and cloud (data, once synced via phone, can be accessed at http://Fitbit.com by logging in to your account). According to Fitbit, if it is not synced to the phone, the device only stores 7 days’ worth of data on the device itself (see image 2, from the Versa 4 product manual). So how was 9 days of data extracted on Feb 19?
3. That Nicola’s steps were recorded to the device in 15-minute increments. Fitbit records steps in 5-minute increments in both the app and on the website. This should allow for a higher level of detail as to what happened between 9:18am and when the device stopped recording data (Feb 4). In sum, you can't use some of her Fitbit data to tell us that she unequivocally went in the water—let alone use it to make a postmortem determination—and reject other data as some sort of glitch. We are not getting the full picture here at all. If Google (who owns Fitbit) was somehow magically able to get authorities nine days' worth of data off the device, that could explain that piece. But the rest of it?
Some questions:
1. How did this device exceed regular battery capacity by two days while submerged in 3-degree water?
2. Was the Fitbit (and its owner) ever in the water at all? A Fitbit is not a forensic pathologist.
3. What else could account for the stop in steps, the heart rate elevations after 9:33am on Jan 27, and the heart rate data recorded until Feb 4?
4. Is it just a coincidence that her device stopped recording data on the day the media covered her Fitbit in multiple articles (Feb 4)?