Abstract Body: Objectives: Bed transition times, i.e., the duration to enter and exit a bed, are key indicators of functional recovery in stroke survivors. These transitions provide insights into rehabilitation progress and may trigger intervention needs. Traditional methods rely on video recordings or manual observations, which can be intrusive or impractical for long-term home monitoring. This study evaluates the accuracy of a novel contactless and cameraless sensor in estimating bed transition times for stroke survivors.

Methods: We developed a contactless sensor system with a processing unit to estimate bed entry and exit times. The system uses low-power radio-frequency signals (similar frequency used in 5G smartphones) to capture reflections from a subject's body. It generates a bird's-eye-view of the environment from the reflected signals to identify the bed's geometry within the monitoring area. A time-domain signal cross-correlation module then isolates significant movement changes of the subject from a location to the bed and vice-versa, allowing estimation of bed entry and exit times. We compared the sensor's measurements with standard camera-based methods (Figures).

Results: Ten stroke survivors (mean age 62.4 years, standard deviation or SD = 11.8) with bed entry durations of 4.2 to 22.0 seconds and exit durations of 2.1 to 88.0 s, along with ten control subjects (mean age 31.6 years, SD = 17.7) with entry durations of 2.1 to 6.0 s and exit durations of 2.0 to 5.2 s, were evaluated. For stroke survivors, the contactless sensor showed a high correlation (R = 0.99) with camera-based measurements for both entry and exit durations, with mean errors of 0.93 ± 0.79 s for entry and 1.86 ± 1.61 s for exit. For control subjects, the correlation was moderate for entry (R = 0.82) and low for exit (R = 0.57), yet the sensor maintained low error margins, with mean errors of 0.65 ± 0.39 s for entry and 0.79 ± 0.55 s for exit. These results suggest the sensor is well-suited for accurately estimating bed transitions despite variability in durations.

Conclusions: This contactless sensor offers a viable alternative to camera-based monitoring for tracking bed transition times in stroke survivors. Its accuracy in detecting varying entry and exit durations highlights its potential for integration into home-based stroke rehabilitation. This sensor could improve stroke recovery management by providing continuous, real-time patient mobility data, enabling proactive interventions.