Researchers at Dartmouth College and Clemson University are working on a new type of sensor that can better help people to understand how much they eat. While there are plenty of apps and services designed to assist in calorie and food intake counting, they normally do rely on the user to input the details to begin with. Which if nothing else, can be quite a time-consuming process. This is where this particular sensor comes in, as the idea is that it is able to calculate how much you eat – by tracking how much you chew.
At the moment the sensor is in the very earliest stages of development and is unlikely wind up as an on the shelf product in the near-term. At present, it is just the sum of research being conducted by the two academic institutions and has only drawn on a small sample pool to highlight the possible benefits of such a sensor. Although, in spite of that small pool of participants, the research paper notes that the "preliminary results show that the system we implemented can detect eating with an accuracy exceeding 90.9-percent." There are some fundamental issues though that will need to be worked through before it becomes a viable product. One being the sensor is unable to tell exactly what type of food a person is eating, or even determine the difference between different crunch levels of food. Likewise, the system is also prone to picking up false positives. For instance, when someone is chewing something but not swallowing, the system will designate the action as eating – which might not be the case. On the other hand, the system does also currently fail to recognize drinking in the same way as eating. In terms of calorie counting purposes, this could lead to downplaying the amount of calories that are being consumed by an individual.
As for the actual sensor, the research paper notes that there has been previous attempts at creating such a sensor, and these are typically placed at a point which is considered to be good for accurate results. Namely, inside the ear canal or against the throat. Although the paper notes that neither are user friendly positions. In contrast, the reports also notes that while typical sensors that are worn on the wrist are more user friendly, they often do not lead to accurate enough results. As such, this sensor looks to deal with both issues by being placed behind the ear. A location which is unlikely to affect the wearer, but also physically close enough to the chewing area to pick up sound and motion data. Going forward, the researchers explain that they are working on "a wearable system that can capture, process, and classify sensor data to detect eating in real-time." While such a system may never prove to be totally reliable, the one interesting implication for this sort of wearable is that it may be able to provide feedback to users on how much they chew, with that information able to be directly extracted and used. For instance, feedback on how much someone snacks compared to how many square meals they consume on a daily basis – not quite what they eat, but their eating habits in general.