Researchers Develop Device Able to Track Food Consumption

Recently, researchers from Carnegie Mellon University developed a wearable device that can track the amount of food being eaten, and the sensors used by the devices are commonly found in smartphones. How are commonly found sensors able to expand the capabilities of everyday devices, what did the researchers demonstrate, and why are they asking for more specific sensors?

How sensors in smartphones are enabling a broad range applications

The development of the smartphone marked an important milestone in human history. Even though powerful computational devices had already existed prior to the invention of the smartphone and the internet was already an established piece of infrastructure, the smartphone brought computation and the internet into our pockets. 

Instead of having to find an internet cafe to send an email or make a mental note to compose the email when returning home, the introduction of the smartphone effectively connected humans to the internet (and, by extension, any cloud resource) such that most computing tasks can be done anywhere at any time.

Fast forward to 2022, and smartphones have been upgraded countless times with new screen technologies, better performance, and more capabilities. But smartphones have also seen the integration of numerous sensor technologies, including GPS, accelerometers, cameras, microphones, gyroscopes, magnetism, and even LiDAR, all of which are used to improve the performance of smartphones.

But while many developers are using these sensors at face value (i.e. 3D scanning using LiDAR and drop detection using accelerometers), others have started to combine such sensor readings with advanced neural networks and deep learning models to develop applications not previously considered.

For example, the camera and flash on a smartphone can be used as a heartbeat monitor by simultaneously placing the finger over the light and camera. As the heart pumps blood, the increase in blood pressure changes the colour of the finger (albeit very slight). Thus, a simple app that looks for this change in the camera can detect heartbeats.

Another potential use of smartphone sensor data is improving sleep patterns. It is believed that the best time to wake up is at the end of a 90-minute sleep cycle, and such sleep cycles can be easily determined by the amount of movement during sleep. Therefore, a smartphone placed on the bed will detect movement during sleep (via accelerometers) and can then wake the user up during the transition phase between two sleep cycles (those who follow this method claim that a 90-minute sleep is better than a 7-hour sleep if woken up during the middle of a cycle). 

Overall, numerous developers are looking for new ways to use smartphones to improve everyday life, especially in the field of healthcare. 

Researchers develop a wearable device to measure eating habits

Recently, researchers from Carnegie Mellon University announced their development of a wearable device that can track eating habits called FitByte. The device is mounted into the frame of a pair of glasses and includes a camera, gyroscope, accelerometer, and proximity sensor.

The camera is targeted onto the face of the wearer to visually see what is being eaten, while the gyroscope is used for monitoring jaw movements (i.e. chewing). The accelerometers monitor swallowing via small vibrations, and the proximity sensors detect gestures from hands as they approach the mouth. Overall, the device has demonstrated its ability to track the amount of food eaten by the wearer and, by extension, provide important metrics to those looking to lose weight. 

This use of technology is beneficial for those trying to diet for multiple reasons. Firstly, those on diets are often required to write food diaries to track what is being eaten and how much, but this is not always practical. Even if a food diary is kept, actually recording how much was eaten can also be challenging as food thrown away may not be considered. 

Secondly, using a smart device to track eating habits can be highly accurate by timestamping food consumption by the second (something a food diary may struggle to do). 

Thirdly, it may even be possible to tie visual data from food being eaten to an image recognition AI that can make an educated guess about the number of calories consumed. This data can then be subtracted from the estimated number of calories burned to produce an overall net gain/loss figure.

Why are researchers asking for more “specific” sensors?

Even though great strides have been made in smartphone sensors and how they can be used in medical applications, the researchers from Carnegie Mellon University stated how they want smartphone manufacturers in the future to integrate more specialised sensors such as blood glucose monitors and sweat sensors. According to the researchers, integrating such sensors into smartphones enables them to become extremely powerful medical platforms while keeping their costs extremely low (compared to mainstream medical equipment). 

This can allow real-time health monitoring to become a reality whereby smartphone users are monitored 24/7 by their devices. Such medical data can then be analysed by machine learning algorithms to identify potential issues early on, which could be a significant step toward increasing general well-being and extending life expectancy.

Of course, smartphones are only so large and are limited to non-invasive sensing methods, which restricts what they can do. But that doesn’t mean future smartphones can’t have microneedle arrays placed into the back for convenient measurements or contact pads that can use electrochemical sensing to detect specific compounds in a drop of blood. In fact, such a contact sensor could even be potentially utilised to analyse breath for traces of infectious diseases such as COVID. 

Overall, smartphones have proven themselves to be extremely powerful sensor platforms, and even though researchers are asking for more, the medical abilities of current devices are already far beyond what anyone would have imagined 10 years ago.