المتخصصون في المجالات الطبية

Oct. 10, 2025

Wearable sleep assessment tools equipped with accelerometers have been used to study sleep in people with sleep disorders and chronic pain. These at-home, noninvasive sleep assessment tools can collect high-resolution, objective sleep data that correlate well with polysomnography (PSG), a widely accepted method for accurately evaluating sleep.

In an article published in PLOS One in 2025, Paul M. Scholten, M.D., and colleagues from Mayo Clinic in Minnesota and Florida discuss some challenges associated with currently available wearable sleep assessment tools. They also describe their efforts to optimize a wrist-worn accelerometer-based, automated sleep detection methodology for patients experiencing chronic pain. Dr. Scholten is a physiatrist with subspecialty training in pain medicine who serves as the practice chair of research for Mayo Clinic's Spine Center in Rochester, Minnesota.

Dr. Scholten and co-authors explain that although this evolving technology may help overcome some of the limitations associated with PSG and self-reported questionnaires, using accelerometry for automated sleep detection is inherently difficult. Accurate extraction of relevant measures requires significant preprocessing and careful selection of appropriate algorithms. The co-authors note that the Tudor-Locke (TL) algorithm can provide investigators with useful sleep quality measures derived from the accelerometry data recorded by the device. However, the accuracy of this algorithm depends on its parameter settings. And while the TL algorithm has been validated for children, its suitability for patients with chronic pain and other populations likely to exhibit abnormal sleep characteristics remains unclear.

"There has been a recent, rapid increase in the utilization of this type of technology in the research world, and its use in the clinical setting is gaining traction," explains Dr. Scholten. "Our research team's goal was to establish best practices and validate our methods for the analysis of this type of data in a very specific clinical setting. This is an important and necessary step to ensure valid results in our team's forthcoming work and to facilitate future potential clinical interpretation of these results."

Study methods

A cohort of 16 patients with high-impact chronic pain participated in a one-week at-home observation before beginning an interdisciplinary pain management program. Participants were asked to keep a sleep diary and wear an activity monitor equipped with an accelerometer supplied by the research team to record baseline sleep routines for at least five nights. The researchers used the TL sleep detection algorithm to derive sleep quality measures from the accelerometry data recorded by the device.

Because this algorithm had not been validated for patients with chronic pain, the researchers conducted a sensitivity analysis of the algorithm's parameters. This analysis helped the researchers develop a new set of parameters that maximized agreement between sleep periods identified by the algorithm and sleep periods identified by participants' sleep logs. The researchers then compared sleep measures derived using the optimized parameters with sleep measures derived using the default parameters.

Results

Overall, the study findings suggest that an optimized parameter set should be used for assessing pain in people with chronic pain. Specific findings supporting that conclusion include the following:

• Data collected using the algorithm's default parameter set achieved a mean sleep detection agreement with participants' sleep logs of 50%.
• Data collected using the algorithm's optimized parameter set achieved a mean sleep detection agreement with participants' sleep logs of 67%.
• There were statistically significant differences between sleep measures from the optimal and default parameter sets (P < 0.001).

Conclusions and next steps

Dr. Scholten and co-investigators state that caution is required when applying the TL algorithm to perform automated sleep detection extracted from accelerometry data in people with chronic pain and other patient populations outside of the algorithm's original validated scope.

The researchers note that the optimized parameters should provide a more reliable starting point for identifying sleep periods. This should enhance the accuracy of the sleep metrics derived by the algorithm and the assessment of patient outcomes and intervention effectiveness related to sleep in patients with chronic pain.

"I think the most significant outcome of this study was that we now have the confidence to interpret data about the quality and duration of patients' sleep without the need to have a diary from the patient," says Dr. Scholten. "Traditionally, the need for these patient-reported times has, in a lot of ways, defeated the purpose of using this technology to collect data. But now that barrier has been removed."

Incorporating this knowledge into a chronic pain management program setting is a logical next step. "We plan to apply what we have learned in a group of patients participating in an interdisciplinary pain management program, assessing how both sleep and activity patterns change after treatment. In the future, I suspect this information will also help us deliver more individualized care," says Dr. Scholten.

For more information

Faust L, et al. Optimizing an automated sleep detection algorithm using wrist-worn accelerometer data for individuals with chronic pain. PLOS One. 2025;20:e0319348.

Refer a patient to Mayo Clinic.