A pragmatic assessment of the influence of data processing on Bluetooth oximetry in pulmonary rehabilitation.

Longer running averages (15-s and 30-s) narrowed limits of agreement for pulse oxygen saturation and heart rate during resting conditions but not during exercising conditions.

• Narrowing of limits of agreements for pulsed oxygen saturation and heart rate were seen with 15 and 30-s running averages for the resting condition only.
• No apparent differences in limits of agreement were observed for exercising conditions across data processing methods.
• Agreement was assessed using Bland-Altman plots comparing the Bluetooth oximeter to a wired oximeter used simultaneously.

Overall failure rates were significantly reduced with 15-s and 30-s running averages compared to 3-s running averages.

• Failure rates for 15-s and 30-s running averages were 36% and 35%, respectively.
• Failure rate for 3-s running averages was 39%.
• Differences in failure rates were statistically significant at adjusted P < 0.05 using Cochran Q tests.

The study recruited 14 individuals from ambulatory and home-based pulmonary rehabilitation programs to evaluate Bluetooth oximeter performance.

• Participants were recruited from the Alfred Hospital (Melbourne, Australia).
• Oximetry was collected during a single rehabilitation session.
• Data collection included resting (2 min) and one or two occasions of exercising (5 min of walking and/or cycling, or Modified Incremental Step Test).
• Two devices were used simultaneously: one Bluetooth oximeter and one wired oximeter.

The Bluetooth oximeter collected data at 1 Hz and was processed using three different running average windows for comparison.

• Data from the Bluetooth oximeter was processed using 3-s, 15-s, and 30-s running averages.
• The high sampling rate of the Bluetooth oximeter (1 Hz) enabled post-hoc application of multiple averaging methods.
• Agreement with the wired oximeter was assessed using Bland-Altman plots for each processing method.

The performance of Bluetooth oximeters is negatively affected by exercise, motivating investigation of data processing approaches to improve data quality.

• Bluetooth oximeters are increasingly available as smart devices that may improve access to pulmonary rehabilitation.
• High sampling rates of these devices were identified as enabling the use of running averages to reduce artifacts.
• The study was designed to explore how data processing influences performance specifically during pulmonary rehabilitation activities.