Summary
A medRxiv preprint describes an open-source framework that synchronized 6,384 hours of chronic deep brain stimulation recordings with wearable data from three people with obsessive-compulsive disorder.
Researchers have described a software framework for combining long-term neural recordings from sensing-enabled deep brain stimulation (DBS) systems with behavioural data from wearable devices. In a medRxiv preprint, the team reports synchronising 6,384 hours of intracranial recordings and wearable measurements collected over months from three participants with obsessive-compulsive disorder (OCD).
The framework is intended to make neural data easier to interpret in everyday settings. It combines automated wearable-data ingestion, handling of neural recording artefacts, time alignment in defined epochs and production of analysis-ready datasets. The authors describe the software as publicly available and reproducible.
Combining brain signals with everyday behaviour
Sensing-enabled DBS systems can record neural activity while a device is implanted for stimulation. Those recordings show patterns of brain activity, but their interpretation depends on knowing what was happening at the same time. A wearable can provide that surrounding behavioural and physiological context, including sleep-wake state, physical activity and autonomic measurements such as signals related to the body’s involuntary nervous-system activity.
In this evaluation, the researchers paired chronic DBS recordings from Medtronic Percept devices with data from Oura Rings. The resulting datasets also included DBS parameters, allowing neural recordings to be considered alongside device settings and periods of daily activity or sleep.
The software pipeline addresses several practical steps in this process. It imports wearable data, manages artefacts that can affect neural recordings, and aligns measurements from devices that collect data on different clocks and schedules. The output is a multimodal dataset: a dataset that contains several types of measurement tied to a common timeline.
A timing test supported alignment of the data
A key technical issue is clock drift, in which one device’s clock gradually moves ahead of or behind another’s. In benchtop testing, the researchers found that the clock in a Medtronic Percept system shifted by 12 seconds relative to network time over one week.
The chronic neural recordings had a sampling interval of 10 minutes. The measured drift was substantially shorter than that interval, and the authors report that this supported reliable alignment with the wearable-derived data. The framework therefore allowed the team to join long-duration neural and wearable recordings without treating their timestamps as perfectly identical.
The study’s main result was the creation and synchronization of these longitudinal neurobehavioral datasets, rather than a measured improvement in OCD symptoms or a new clinical treatment effect. The source reports that the work received approval from the Baylor College of Medicine institutional review board under protocol H-43144 and that participant consent was obtained.
Why the framework matters for neurotechnology research
Neural signals recorded during controlled laboratory sessions can be difficult to relate to the conditions in which symptoms and behaviour occur over weeks or months. Pairing chronic recordings with sleep, movement and physiological context can help researchers examine how neural patterns vary across naturalistic settings.
The authors propose that the framework can reduce technical barriers to longitudinal neurobehavioral studies and provide a foundation for biomarker discovery. A biomarker in this context would be a measurable neural or physiological pattern associated with a state, symptom or outcome. Establishing a clinically useful biomarker would require further studies; this preprint reports the infrastructure for collecting and aligning the relevant measurements.
The evaluation was limited to three participants with OCD and used a particular combination of DBS and wearable systems. It was a technical framework demonstration, with the central outcome being synchronized data generation. The results therefore provide a basis for future research into behaviourally annotated neural recordings rather than evidence that the framework itself changes treatment or diagnoses OCD.