Smart-Motion on Fitbit Sense [Development Stopped]

Work on this project has been stopped due to Fitbit discontinuing the Fitbit Sense. We leave the below published for those interested in understanding what we think is a good design for a cryonics alarm using a fitness tracking device. Our efforts are currently now focused on building a monitoring system from scratch.


For about two years Nikki analyzed, tested and prototyped possible solutions. A few months ago, the Fitbit Sense got an update that fixed a bug where vibrating due to notifications interfered with the accelerometer and gyroscope. This paved the way for a good cryonics alarm solution that utilizes pulse and motion detection. Due to the sophisticated usage of heart rate and motion sensors by this solution it is very unlikely that if the wearer dies, the watch app doesn’t detect it (presumed that the wearer handles the watch in accordance with our guide lines). This article is about the above-mentioned solution which is currently in the Alpha phase of development.

General Information About the System

The solution consists of three components: a Fitbit Sense watch app, a Fitbit smartphone app (available for iOS and Android) and a website.

The main functionality of the website is to add/edit contacts and configuration in which order and how these should be contacted when the alarm is active. Besides the order in which the contacts will get contacted, it is also for each contact individually configurable by which medium, and in which media order, she/he gets notified. Supported media are phone call, text (SMS) and email.

Our website interface for editing a contact

When a contact gets notified by our automated system due to the triggering of the alarm, the website offers them to download necessary/useful documents for handling the case. These include among other things their cryonics contract and will. For the user the website also offers the possibility to change settings how the system should behave for them and their contacts. For example, change the time span after which the next medium is used to notify a contact about a triggered alarm.

Disconnection screen

The mobile app mainly functions as internet access for the watch app. The watch app communicates with the mobile app via Bluetooth and can send/receive data to/from our cloud software. To reduce the time in which the wearer's death would not be detected by our system, the watch app warns the user when there is no Bluetooth connection to the smartphone for longer than a user defined custom number of minutes. The warning happens via vibrating and displaying "Disconnected" on the screen. The default time we suggest before being notified is 5-10 minutes, as Bluetooth will sometimes disconnect and then reconnect even when in range.

The disconnection screen (pictured) also displays the amount of time the watch has been disconnected so far. This is to help the user get a sense of their issues with Bluetooth so they can understand when they are not being monitored. We also keep track of this data and display the disconnection times to the user in a dashboard in the user's profile.

The Escalation Process

When the Fitbit watch app recognizes no pulse, it immediately triggers by displaying a message that it is in the triggered status and by vibrating. In this state the user can confirm via a button press on the watch that it is a false alarm.

Monitoring system in triggered state

If the button to confirm that it is a false alarm is not pressed within 5 minutes, the watch app switches to the alarm state. When this state change occurs, our system tries to contact the user via one or more media (configurable). If she/he then doesn't confirm that it is a false alarm within 5 minutes (e.g. when called via a button press), the system tries to reach out to the first contact in the contacts list. The notified contact can confirm that she/he will handle the case by pressing a button when being called or by answering to a text (SMS) / email. In order to be able to find the wearer of the watch as fast as possible, the contact gets informed about the latest GPS coordinates of the watch. Besides the GPS coordinates the contact also gets a website link to google maps which already contains the GPS coordinates. Additionally, the contact receives another website link where they can download a bunch of useful/necessary documents to pave the way for a fast cryopreservation. If the first contact doesn't confirm that she/he will handle the case within 5 minutes, the next contact gets notified. If the second contact also doesn't confirm that she/he will handle the case within 5 minutes, the next contact gets notified, etc.

Monitoring system in activated state

False Alarms, Optimizations and Comparisons to Other Systems

We estimate it as very unlikely that one of the user’s contacts get notified without there being a real emergency (considering the user acts according to our guidelines). The probability is estimated to be very low because we designed our sensor data evaluating software component very carefully in order to reduce the number of false alarms, which we tested extensively. Moreover, the user gets notified via watch vibrating and phone call that the watch got triggered and has several minutes to mark it as a false alarm. To optimize the average death detection time, we developed a movement data evaluating algorithm which is tailored for the accelerometer and gyroscope sensor data of the Fitbit Sense watch.

Wrist movement at night - during sleep still small movements

Based on the usual behavior of the user, the algorithm dynamically adjusts the number of minutes which need to pass without motion detection to trigger the alarm. For example, when the user is not sleeping, the number of minutes can be reduced a lot. This is due to the fact that humans move their arms more frequently while awake.

During the day, the user moves their wrist 1000s of times, and so it's very easy to get a clear signal of whether or not they are ok. The user may take naps during the day, but the total time without moving during a nap is similar to that seen during the night. Also, it is known that as we age we sleep less soundly and have more movement during the night. The prototype and proof of concept on this movement component is based off of gathering movement data on ourselves.

Due to the fact that the pulse sensor in some circumstances detects the missing of a heart rate in less than one minute, our solution has the same best-case detection time as the cryonics alarm/monitoring systems we reviewed (Tomorrow Bio monitoring app, CI Check-In app, Alcor Check-In app, Gene Shaver's Sensor app). The system also responds in less than 1 min when the user's HR passes through a high or low threshold.

Regarding the average-case our solution is, with 45 minutes, the best, followed by the Tomorrow Bio monitoring app with 9 hours. Even in the worst-case our solution detects death within two hours, which is better than every other of the solutions we reviewed. For more details regarding how these times were calculated view our Reviews articles.

How We Compare: Best, Worst and Average Alarm Times


Passive monitoring solutions (our solution, the Tomorrow Bio monitoring app [when check-in functionality is disabled] and Gene Shaver's Sensor app) obviously have a big comfort advantage over solutions which require a frequent check-in. One noteworthy disadvantage of our solution is that it currently only works with the Fitbit Sense. We intend to support more devices in the future, since our website and cloud software works with basically any device.

Thanks to death detection via two different signals (motion and pulse) and several optimizations regarding algorithms for data analytics we managed to reduce the average death recognition time to under 45 minutes. This is a much better average-case than any other cryonics alarm/monitoring solution we analyzed/tested. The worst-case death detection time of our solution (two hours) is also better than any other tested solution. Overall our system would be a great addition to the set of cryonics alarm/monitoring solutions currently available, especially for people who valuable a substantially faster death detection.

View the system demo video: