Author(s): DECORTE, R., SLEMBROUCK, M., VERSTOCKT, S., Institution: GHENT UNIVERSITY, Country: BELGIUM, Abstract-ID: 1260

INTRODUCTION:
Transferring laboratory tests and data collection procedures to large-scale field tests presents several challenges due to the inherent differences in controlled laboratory environments versus dynamic field conditions. The extra manual labor required for following up a large-scale study and some logistical concerns such as equipment portability and data collection in remote locations further complicate the transition from lab to field. For a performance and health related study of military recruits, we are developing a continuous monitoring setup using consumer available smartwatches that requires minimal effort from the study operators to obtain the relevant data.
METHODS:
We developed an end-to-end pipeline which monitors study participants 24/7 using a Garmin smartwatch. Health parameters measured by multiple Garmin smartwatches are transmitted to a single phone using an operator sync (participants are not required to upload their own data). The proposed solution is tailored towards complex environments where the applicability must be non-invasive to not disrupt daily activities and has to be decoupled from external servers to ensure confidentiality. As such, study participants of the military recruits can be monitored using a smart watch with GPS capabilities, even during multi-day tactical exercises in the field or during prolonged periods without connectivity. Based on the weekly itinerary, the watch uses different monitoring modes and sampling rates to extend battery life while providing a wide range of (optional) parameters such as heart rate, GPS, movement dynamics, accelerometer, etc. This is linked together with other data sources such as itinerary metadata, test battery results, injuries and more.
RESULTS:
The proposed system has been used in an 8-week-long study. During this period, it was used to collect specific GPS based activities (map reading exercises, runs at aerobic and anaerobic thresholds, etc.) and various monitoring parameters (heartrate, steps, sleep quality, stress, etc.). Due to the operator synchronization mechanism, it is easy to deploy such a system in a field test (short or long deployments). The recorded parameters are immediately available when the operator chooses to synchronize the devices. The uploaded data is then automatically linked to the corresponding profile in our central platform and the annotations from other data sources are applied for improved analysis.
CONCLUSION:
First results show the feasibility of the proposed approach. However, some remaining issues still need to be resolved in regard to the Garmin Health SDK, which at the moment fails to transfer all the available data from the watch to the phone if a large number of watches are synchronized simultaneously. The valorization potential of the proposed set-up is high as it could also be used in other contexts. E.g. research laboratories that can quickly equip a group of people for their testing without having to manually process all the watches afterwards.