Goerss, Doreen;
Köhler, Stefanie;
Haufschild, Martin;
Bader, Sebastian;
Kirste, Thomas;
Teipel, Stefan J.
Sensor‐based activity and state recognition in dementia patients in stationary care as basis for situation‐aware assistive devices : Neuropsychiatry and behavioral neurology/assessment/measurement of neuropsychiatric/behavioral and psychological symptoms
You can manage bookmarks using lists, please log in to your user account for this.
Media type:
E-Article
Title:
Sensor‐based activity and state recognition in dementia patients in stationary care as basis for situation‐aware assistive devices : Neuropsychiatry and behavioral neurology/assessment/measurement of neuropsychiatric/behavioral and psychological symptoms
:
Neuropsychiatry and behavioral neurology/assessment/measurement of neuropsychiatric/behavioral and psychological symptoms
Contributor:
Goerss, Doreen;
Köhler, Stefanie;
Haufschild, Martin;
Bader, Sebastian;
Kirste, Thomas;
Teipel, Stefan J.
Description:
<jats:title>Abstract</jats:title><jats:sec><jats:title>Background</jats:title><jats:p>Wearable sensors to monitor activity are easily available nowadays. Intelligent assistive technologies bear potential to offer innovative solutions that increase safety and facilitate independent activities in persons with dementia (PwD). It is important to incorporate knowledge about user‘s current state (e.g. cognitive and physical abilities) and situation to generate suitable interventions and provide meaningful assistance for PwD.</jats:p></jats:sec><jats:sec><jats:title>Method</jats:title><jats:p>Within the “SAMi”‐project, we set up an (ongoing) field study with PwD living in a nursing home. Participants wear smartwatches to measure activity (accelerometry) and enable indoor positioning via Bluetooth. In parallel, subjects' behavior is annotated in real time by trained observers using an elaborated scheme that contains activity, posture and place of the person, as well as disorientation, falls or request for help. The annotation scheme and possible intervention domains were derived from interviews with different groups of stakeholders. With the field study, we aim to identify in real time if a patient is at need of support and to select the most promising intervention in a given situation. Therefore, we plan to train algorithms on the synchronized sensor‐data and examine the accuracy of situation‐classification.</jats:p></jats:sec><jats:sec><jats:title>Result</jats:title><jats:p>To date, we included nine PwD in the age of 76‐96 years in moderate to severe stages of dementia (MMSE 5‐18). We gained >183 hours of observational data, accompanied by sensor data. We observed 134 events of disorientation, 95 events of caregiver support but only 3 requests for help. Ca. 79% of the annotated time patients wore the smartwatch. The majority of patients accepted the device (despite advanced cognitive decline). Setting up the technical infrastructure for the indoor positioning was challenging, for example due to restricted access to power points.</jats:p></jats:sec><jats:sec><jats:title>Conclusion</jats:title><jats:p>Analysis of data from wearable and ambient sensors bears potential to build a situation‐aware assistive device and is feasible to acquire in the setting of dementia patients living in nursing homes. A large number of situations where need for help became obvious was matched only by very few instances of active help seeking by the patients. This emphasizes the necessity for automated sensor‐based recognition of such situations.</jats:p></jats:sec>