Digital Endpoints: Sleep

Real-World Assessment of Sleep-Related Outcomes

Real-World Assessment of Sleep-Related Outcomes

Sleep is related to all aspects of physical and mental health. There is a cyclical relationship between chronic sleep disruption and disease. Poor sleep is both a risk factor for and the possible effect of cardiovascular, metabolic, neurodegenerative, and mental health disease.

Typical sleep-related outcome measures are interrelated and include total duration, timing, latency, efficiency, wake-after-sleep-onset, and fragmentation. Like all real-world outcomes, these measures' accuracy and precision are primarily dependent on sensor modality and the patient population's characteristics. Most automated sleep detection algorithms were developed in healthy people and prone to large measurement errors when applied to populations with disordered sleep.

VivoSense Software allows the ability to customize standard sleep endpoints to manage artifactual data, adjust standard thresholds, provide expert human over-read, and use machine learning and AI to improve standard outputs. Novel sleep-related outcomes can also achieve results relevant to specific clinical disease states or those not commonly assessed.

Wearable Technology

Sleep is a dynamic physiologic state that impacts virtually all physiological processes. Sleep health is a multi-dimensional phenomenon defined by five key characteristics: duration, timing, quality, daytime alertness, and absence of a sleep disorder. Because of this complexity, sleep-related outcomes are traditionally measured in a laboratory setting using a multi-sensor polysomnography (PSG) system.

PSG's individual sensor modalities are deployed within their wearable form-factor to measure sleep's specific components in the real world.

EEG: Wearable electrode sensors are used to measure the brain's electrical activity and produce a high-resolution time series representation of voltage fluctuations. The resulting electroencephalogram (EEG) can then be used to identify distinct brain waves associated with sleep's different components. The need for electrodes to be placed around the scalp limit EEG's use in the real world. Relatively low profile and minimally burdensome technologies exist to measure sleep components that are not otherwise available from the more commonly used real-world sleep solutions.

Accelerometer: Using accelerometry to identify periods of sleep vs. wake is based on the premise that sleep is characterized by minimal or no movement periods. Despite the obvious limitations of using lack of movement to distinguish sleep from wake, including the fact individuals can remain quite still despite being fully awake, wrist accelerometry remains the most common means to assess components of sleep in real-world settings.

Newer methods and technologies have begun to simultaneously incorporate behavioral information measured from actigraphy with physiologic signals from other sensors, such as heart rate, body temperature, and EEG, to improve the accuracy of sleep-related outcomes.

Outcome Measures

Total Sleep Duration

Sensor Modalities: Accelerometer, EEG

Clinical Use Examples: Monitoring biomarker for assessing drug related sleep disruption side-effects in patients taking dopamine agonists, such as in Parkinson's disease.

Sleep Timing and Circadian Rhythm

Sensor Modalities: Accelerometer, EEG

Clinical Use Example: Response biomarker for clinical trials assessing the side-effects of chemotherapy on circadian activity and quality of life.

Sleep Fragmentation

Sensor Modalities: Accelerometer, EEG

Clinical Use Example: Predictive biomarker for over-the-counter analgesics.

Sleep Efficiency

Sensor Modalities: Accelerometer, EEG

Clinical Use Example: Response biomarker for obstructive sleep apnea efficacy trials.

Wake After Sleep Onset (WASO)

Sensor Modalities: Accelerometer, EEG

Clinical Use Examples: Monitoring biomarker for assessing disease progression in patients with Alzheimer's disease.

Talk to One of Our Data Scientists