Isolated REM sleep behavior disorder (iRBD) is one of the strongest early warning signs in neurology. More than 80% of people diagnosed with iRBD will eventually develop a neurodegenerative disease — typically Parkinson's disease, dementia with Lewy bodies, or multiple system atrophy — within 14 years. Catching it early opens a window for neuroprotective interventions that may slow or prevent irreversible damage.
The problem is that diagnosing iRBD currently requires an overnight stay in a sleep laboratory for video-polysomnography (vPSG), a resource-intensive procedure that is difficult to access and captures only a single night's data. A new study published in npj Digital Medicine offers a potential alternative that patients can use at home.
How the Study Worked
Researchers at Tel Aviv University enrolled 73 participants — 15 with confirmed iRBD and 58 age-matched controls without the condition. All participants first underwent standard video-polysomnography in a sleep laboratory to establish ground truth diagnoses.
Each participant then took home a small inertial measurement unit (IMU) — a motion sensor roughly the size of a matchbox — and wore it on their lower back for six consecutive nights while sleeping normally in their own beds. The device recorded movement patterns throughout each night.
The researchers extracted mobility features from the sensor data and trained machine learning models to distinguish iRBD patients from healthy controls based on their nocturnal movement signatures.
High Sensitivity, Improving With More Nights
The best-performing model, a support vector machine (SVM) classifier, achieved 93.33% sensitivity and 72.41% specificity, with an overall accuracy of 76.71% and a balanced accuracy of 82.87%.
The 93% sensitivity is the critical metric for a screening tool: it means the system correctly identified nearly all iRBD patients, minimizing the chance of a missed diagnosis. The moderate specificity — meaning some healthy sleepers were flagged as potential cases — is an acceptable tradeoff in a screening context, where flagged individuals would proceed to confirmatory laboratory testing.
A key finding was that performance improved as more nights of data were included, plateauing at approximately five nights. This suggests that the variability inherent in sleep — particularly the intermittent nature of REM sleep motor episodes — requires multi-night sampling to capture reliably.
Why the Lower Back
The choice of sensor placement was deliberate. REM sleep behavior disorder is characterized by the loss of the normal muscle paralysis (atonia) that occurs during REM sleep, causing patients to physically act out their dreams — punching, kicking, flailing, or vocalizing during sleep. While wrist-worn accelerometers are common in sleep research, the lower back captures whole-body trunk movements that are more representative of the large motor behaviors characteristic of RBD episodes.
Previous research has validated lumbar-mounted sensors for movement analysis in other neurological conditions, including Parkinson's disease gait assessment. The Tel Aviv team's work extends this approach to sleep-based screening.
Filling a Diagnostic Gap
The clinical need is substantial. RBD affects an estimated 1–2% of the general population, but the vast majority of cases remain undiagnosed. Many patients are unaware of their nighttime behaviors — it is often a bed partner who first notices the thrashing, talking, or violent movements during sleep.
Even when patients or their partners report symptoms, accessing video-polysomnography can involve months-long wait times. The limited number of sleep laboratories worldwide creates a bottleneck that delays diagnosis, particularly for patients in rural areas or countries with fewer sleep medicine resources.
A home-based screening tool could serve as a first-stage filter, identifying high-probability cases for expedited laboratory confirmation while sparing low-risk individuals from unnecessary testing.
Limitations and Next Steps
The study's sample size — 15 iRBD patients and 58 controls — is small, and the researchers acknowledge that validation in larger, more diverse populations is needed before the approach can be used clinically. The sensor also cannot replace polysomnography for a definitive diagnosis; it is designed as a screening tool to identify who should be referred for further evaluation.
The team plans to conduct multi-center validation studies and explore whether the system can also track disease progression over time, potentially serving as an objective biomarker for clinical trials of neuroprotective therapies.
What This Means for Patients
If you or a bed partner have noticed unusual movements during sleep — acting out dreams, talking, yelling, or limb movements that seem purposeful rather than the occasional twitch — speak with a sleep specialist or neurologist. These behaviors may be benign, but they can also be an early signal worth investigating.
Home-based wearable screening for RBD is not yet commercially available, but this research suggests it is technologically feasible. As validation studies progress, patients may eventually be able to undergo initial screening from their own beds, dramatically reducing the time between first suspicion and a confirmed diagnosis.
The full study is available in npj Digital Medicine.
