Patient's reaction to smartwatch based interventions in a controlled environment: preliminary results from the SAMi intervention study

Background: Assistive technologies show promising features to support people with cognitive impairment in daily life, e.g. mobile sensors for falls detection or getting lost. On the other hand, smartwatches bear potential to not only monitor but also interact with the user. While sensor based monitoring in dementia care is already part of many projects, direct interaction of people with cognitive impairment with smartwatches is little investigated. Method: Based on qualitative studies, we designed interventions for a usual consumer smartwatch. We conceptualized two different tasks: A) drinking water and B) circling bells on a worksheet. Moreover, we implemented two different modes of intervention-intensity affecting vibration patterns, alarm sounds, text sizes, display times or appearance of images and animations. We observed patients’ reactions to interventions A and B remotely via cameras. In case of failure, interventions were repeated up to three times. Here we present data of the interim analysis after completion of n=20 patients that were equally assigned either to mode 1 (regular interventions) or mode 2 (intensive interventions). In addition to observations, participants feedback was obtained with questionnaires. Patients were diagnosed with MCI (n=7) or dementia (n=13) in the local memory clinic. Result: We found it feasible to study patient-smartwatch-interactions with a mixed methods approach. The majority of participants (n=18) were able to successfully complete at least one of two tasks, see Fig. 1. Circling bells (B) was completed only by 11 participants, while drinking water (A) was solved by 16 participants. The regular interventions were successful in 24%, the intensive interventions in 65%. Repetitions did not lead to an improvement of outcome in 13 cases, but in two cases. Reasons for insufficient task completion varied, e.g. patients showed unexpected reactions such as circling the picture on the smartwatch or doing arm circles. One patient fell asleep and did not react at all. Conclusion: Remote observation of patients’ reaction to smartwatch based interventions is technically challenging but offers intriguing insights into human-computer interaction. User-centered design for patients with MCI or dementia differs from the approach for usual smartwatch users and needs to respect special needs and requirements.