Abstract
In healthcare, physical activity can be monitored in two ways: self-monitoring by the patient himself or external monitoring by health professionals. Regarding self-monitoring, wearable activity trackers allow automated passive data collection that educate and motivate patients. Wearing an activity tracker can improve walking time by around 1500 steps per day. However, there are concerns about measurement accuracy (e.g., lack of a common validation protocol or measurement discrepancies between different devices). For external monitoring, many innovative electronic tools are currently used in rheumatology to help support physician time management, to reduce the burden on clinic time, and to prioritize patients who may need further attention. In inflammatory arthritis, such as rheumatoid arthritis, regular monitoring of patients to detect disease flares improves outcomes. In a pilot study applying machine learning to activity tracker steps, we showed that physical activity was strongly linked to disease flares and that patterns of physical activity could be used to predict flares with great accuracy, with a sensitivity and specificity above 95%. Thus, automatic monitoring of steps may lead to improved disease control through potential early identification of disease flares. However, activity trackers have some limitations when applied to rheumatic patients, such as tracker adherence, lack of clarity on long-term effectiveness, or the potential multiplicity of trackers.