Abstract:Sleep timing varies between individuals and can be altered in mental and physical health conditions. Sleep and circadian sleep phenotypes, including circadian rhythm sleep-wake disorders, may be driven by endogenous physiological processes, exogeneous environmental light exposure along with social constraints and behavioural factors. Identifying the relative contributions of these driving factors to different phenotypes is essential for the design of personalised interventions. The timing of the human sleep-wake cycle has been modelled as an interaction of a relaxation oscillator (the sleep homeostat), a stable limit cycle oscillator with a near 24-hour period (the circadian process), man-made light exposure and the natural light-dark cycle generated by the Earth's rotation. However, these models have rarely been used to quantitatively describe sleep at the individual level. Here, we present a new Homeostatic-Circadian-Light model (HCL) which is simpler, more transparent and more computationally efficient than other available models and is designed to run using longitudinal sleep and light exposure data from wearable sensors. We carry out a systematic sensitivity analysis for all model parameters and discuss parameter identifiability. We demonstrate that individual sleep phenotypes in each of 34 older participants (65-83y) can be described by feeding individual participant light exposure patterns into the model and fitting two parameters that capture individual average sleep duration and timing. The fitted parameters describe endogenous drivers of sleep phenotypes. We then quantify exogenous drivers using a novel metric which encodes the circadian phase dependence of the response to light. Combining endogenous and exogeneous drivers better explains individual mean mid-sleep (adjusted R-squared 0.64) than either driver on its own (adjusted R-squared 0.08 and 0.17 respectively). Critically, our model and analysis highlights that different people exhibiting the same sleep phenotype may have different driving factors and opens the door to personalised interventions to regularize sleep-wake timing that are readily implementable with current digital health technology. Disrupted sleep has long term health consequences and affects our day-to-day ability to function physically, mentally and emotionally. But what determines when and how long we sleep? It is well-known that daily light exposure patterns determine the timing of the body clock. However, creating mathematical models that can take realistic light exposure patterns and predict plausible sleep timing has been challenging. Furthermore, nearly all previous studies have focused on developing models for average behaviour, yet sleep timing and duration are highly individual. In this paper, we present a simple model that combines sleep regulatory and circadian processes. The model can take individual light exposure patterns and, by fitting physiologically plausible parameters, describe individual mean sleep timing and duration. We test our model on data collected from 34 older participants. Our modelling approach suggests that some of the participants slept late because of physiological factors, while for other individuals, late sleep was a consequence of their light environment. This approach of combining a model with longitudinal data could be implemented in digital health technology such that your smart watch could tell you not only how you slept last night, but also how to change your light environment to sleep better tomorrow.

A Level Set Kalman Filter Approach to Estimate the Circadian Phase and its Uncertainty from Wearable Data

Efficient assessment of real-world dynamics of circadian rhythms in heart rate and body temperature from wearable data

Predicting circadian phase in community‐dwelling later‐life adults using actigraphy data

Ambulatory assessment of human circadian phase and related sleep disorders from heart rate variability and other non-invasive physiological measurements

Circadian Rhythm Analysis Using Wearable Device Data: Novel Penalized Machine Learning Approach

Predicting circadian misalignment with wearable technology: Validation of wrist-worn actigraphy and photometry in night shift workers.

Hidden Markov models for monitoring circadian rhythmicity in telemetric activity data

Inferring Circadian Rhythms of Cognitive Performance in Everyday Life

Continuous Circadian Phase Estimation Using Adaptive Notch Filter

A Bayesian Circadian Hidden Markov Model to Infer Rest-Activity Rhythms Using 24-hour Actigraphy Data

Towards a Computational Framework for Automated Discovery and Modeling of Biological Rhythms from Wearable Data Streams

Comparative Efficacy of Commercial Wearables for Circadian Rhythm Home Monitoring from Activity, Heart Rate, and Core Body Temperature

Sensor-Based Estimation of Dim Light Melatonin Onset (DLMO) Using Features of Two Time Scales

A novel method to separate circadian from non-circadian masking effects in order to enhance daily circadian timing and amplitude estimation from core body temperature

Circadian Rhythms are Not Captured Equal: Exploring Circadian Metrics Extracted by Different Computational Methods from Smartphone Accelerometer and GPS Sensors in Daily Life Tracking

A real-time, personalized sleep intervention using mathematical modeling and wearable devices

The Combination of Topological Data Analysis and Mathematical Modeling Improves Sleep Stage Prediction From Consumer-Grade Wearables

Automated remote sleep monitoring needs uncertainty quantification

Unsupervised Machine Learning Identifies Latent Ultradian States in Multi-Modal Wearable Sensor Signals

Nonlinear Heart Rate Variability Analysis for Sleep Stage Classification Using Integration of Ballistocardiogram and Apple Watch

Method to determine whether sleep phenotypes are driven by endogenous circadian rhythms or environmental light by combining longitudinal data and personalised mathematical models