Effectiveness of a Batteryless and Wireless Wearable Sensor System for Identifying Bed and Chair Exits in Healthy Older People

Abstract

Aging populations are increasing worldwide and strategies to minimize the impact of falls on older people need to be examined. Falls in hospitals are common and current hospital technological implementations use localized sensors on beds and chairs to alert caregivers of unsupervised patient ambulations; however, such systems have high false alarm rates. We investigate the recognition of bed and chair exits in real-time using a wireless wearable sensor worn by healthy older volunteers. Fourteen healthy older participants joined in supervised trials. They wore a batteryless, lightweight and wireless sensor over their attire and performed a set of broadly scripted activities. We developed a movement monitoring approach for the recognition of bed and chair exits based on a machine learning activity predictor. We investigated the effectiveness of our approach in generating bed and chair exit alerts in two possible clinical deployments (Room 1 and Room 2). The system obtained recall results above 93% (Room 2) and 94% (Room 1) for bed and chair exits, respectively. Precision was >78% and 67%, respectively, while F-score was >84% and 77% for bed and chair exits, respectively. This system has potential for real-time monitoring but further research in the final target population of older people is necessary.

Keywords: bed exits; chair exits; fall prevention; older people; weighted conditional random fields.

Figure 1

Overview of the proposed fall prevention technological intervention. Data collected by the radio frequency identification (RFID) infrastructure is sent in real-time to the bed and chair exit recognition approach stage. Caregivers can be notified via alert messages to assist the electronically identified patient (who) that is performing a bed or chair exit (what), the alert is issued in real-time (when) and the RFID antenna and reader identifiers can indicate the room occupied by the patient (where).

Figure 2

Proposed bed and chair exit recognition approach. Acceleration data from the sensor and additional information such as the received strength of the signal (RSSI) evaluated by the RFID reader are inputs to the recognition approach.

Figure 3

Wearable Wireless Identification and Sensing Platform (W²ISP). (A) older volunteer wearing device on top of clothing; (B) W²ISP parts: (i) circuitry, $18\times 20\times 2$ mm; (ii) flexible antenna, $36\times 85\times 2$ mm; and (iii) isolating silver coated fabric, $230\times 220$ mm; (C) block diagram of W²ISP platform; (D) process of lying on bed to sitting on bed; and (E) process of sitting (in bed or chair) to ambulating.

Figure 4

The two room configurations used in the study. Configuration of equipment with antennas on ceiling level shown as circles and vertical antennas shown as rectangles facing either the bed or chair. (A) Room 1, antenna3 is at ceiling level on top of the bed and the rest of the antennas are on a vertical stand. Antenna2 is inclined towards the chair and antenna1 and antenna4 face front (chest level) towards the bed area; and (B) Room 2, antenna2 and antenna3 are at ceiling level tilted towards the bed and antenna1 is on a vertical stand inclined towards the chair.

Figure 5

Sample of collected sensor data. (A) Raw accelerometer readings along the three axes; and (B) RSSI (received signal strength indicator) received from three antennas in the room and RSSI pattern changes across four activities.

Figure 6

Bed and chair exit recognition. State machine transition model used to recognize a bed or chair exit.

Figure 7

Confusion matrix for data of (a) Room 1 and (b) Room 2. Output of dynamically weighted conditional random field (dWCRF) classifier for labels 1: Sitting-on-bed, 2: Sitting-on-chair, 3: Lying and 4: Ambulating. Results in %.