Semi-Automated Data Labeling for Activity Recognition in Pervasive Healthcare.

Activity recognition, a key component in pervasive healthcare monitoring, relies on classification algorithms that require labeled data of individuals performing the activity of interest to train accurate models. Labeling data can be performed in a lab setting where an individual enacts the activity under controlled conditions. The ubiquity of mobile and wearable sensors allows the collection of large datasets from individuals performing activities in naturalistic conditions. Gathering accurate data labels for activity recognition is typically an expensive and time-consuming process. In this paper we present two novel approaches for semi-automated online data labeling performed by the individual executing the activity of interest. The approaches have been designed to address two of the limitations of self-annotation: (i) The burden on the user performing and annotating the activity, and (ii) the lack of accuracy due to the user labeling the data minutes or hours after the completion of an activity. The first approach is based on the recognition of subtle finger gestures performed in response to a data-labeling query. The second approach focuses on labeling activities that have an auditory manifestation and uses a classifier to have an initial estimation of the activity, and a conversational agent to ask the participant for clarification or for additional data. Both approaches are described, evaluated in controlled experiments to assess their feasibility and their advantages and limitations are discussed. Results show that while both studies have limitations, they achieve 80% to 90% precision.

Naturalistic Enactment to Elicit and Recognize Caregiver State Anxiety.

Caring for people with dementia imposes significant stress on family members and caregivers. Often, these informal caregivers have no coping strategy to deal with these behaviors. Anxiety and stress episodes are often triggered by problematic behaviors exhibited by the person who suffers from dementia. Detecting these behaviors could help them in dealing with them and reduce caregiver burden. However, work on anxiety detection using physiological signals has mostly been done under controlled conditions. In this paper we describe an experiment aimed at inducing anxiety among caregivers of people with dementia under naturalistic conditions. We report an experiment, using the naturalistic enactment technique, in which 10 subjects were asked to care for an older adult who acts as if she experiences dementia. We record physiological signals from the participants (GSR, HR, EEG) during the sessions that lasted for approximately 30 min. We explain how we obtained ground truth from self-report and observation data. We conducted two different tests using the Support Vector Machine technique. We obtained an average precision of 77.8 % and 38.1 % recall when classifying two different possible states: "Anxious" and "Not anxious". Analysis of the data provides evidence that the experiment elicits state anxiety and that it can be detected using wearable sensors. Furthermore, if episodes of problematic behaviors can also be detected, the recognition of anxiety in the caregiver can be improved, leading to the enactment of appropriate interventions to help caregivers cope with anxiety episodes.