Data collection and accuracy for CORE
A common question we get about CORE is about accuracy. Is it really that accurate? How did you test CORE? These are fair questions which we are happy to answer: CORE is tested both in a controlled lab environment and in the field. We primarily use electronic pills as a reference (control) to assess the accuracy of the device and our algorithm.
Our testing protocol
For our testing we recruited several top athletes, most of them are from Switzerland, where we are located. We asked them to wear CORE on the upper body while working out on an indoor bike trainer or treadmill in our lab and while training and competing outdoors. We monitored their core body temperature before, during, and after and also engaged in a variety of different sporting activities under different environmental conditions.
In order to optimize the CORE’s core body temperature calculations for different sport scenarios and to further improve the accuracy of our advanced machine learning algorithm, we tested simultaneously with one of the aforementioned invasive and core body temperature methods to ensure there was a proven accurate baseline. The reference for temperature assessment was the ingestible radio pill (e-pill), which directly measures the temperature of the digestive system as it passes through.
Our testing also followed a set of procedures to replicate the most common situations for athletes. This includes the following scenarios:
- Measurements were conducted at various ambient temperatures and relative humidity, (with a tendency for hot and humid conditions)
- Mixed-age and gender groups were tested
- We ran several series of tests with indoor measurements in a controlled environment and outdoor measurement on hot days
- We concentrated particularly on endurance sports with activity lasting longer than one hour, such as cycling and running. Other sports activities were trialed including swimming, cross-country skiing or team sports.
- The CORE device was attached to a heart rate strap on the lateral side of the ribcage.
- Sports tests were scheduled, to ensure the temperature reference method was measuring reliably and was not be affected by water intake directly.
- Measurement periods were not only during sports activity but also started before the activity and the continued into the recovery period after the activity.
- Heart rate, power, speed and further parameters were assessed to characterize the different sports sessions and to gain further knowledge on how core body temperature affects the sports performance and influences the physiological system.
- The machine learning algorithm was continually trained with new data to optimize its predictive power.
The bulk of the data collection has come from athletes in Switzerland, Dubai, France, Italy, and Spain. This has provided enough data for our R&D team to build an algorithm capable of crunching the data in real time and deliver a continuous and accurate core body temperature measurement. The testing is still ongoing as we continue to verify and improve the CORE device.
As of now, our accuracy is a Mean Absolute Deviation of 0.21 °C when compared to the e-pill. However, our approach is to keep improving our technology so it will remain an ongoing process of gathering more data into the pool to get even better measurements. So to answer to question we are ask the most, “Is it really that accurate? Yes, it’s very accurate!
In addition to internal testing, we also have some references for external studies and projects using the CORE technology.