Application of Machine Learning Methods to Ambulatory Circadian Monitoring (ACM) for Discriminating Sleep and Circadian Disorders

Abstract

The present study proposes a classification model for the differential diagnosis of primary insomnia (PI) and delayed sleep phase disorder (DSPD), applying machine learning methods to circadian parameters obtained from ambulatory circadian monitoring (ACM). Nineteen healthy controls and 242 patients (PI = 184; DSPD = 58) were selected for a retrospective and non-interventional study from an anonymized Circadian Health Database (https://kronowizard.um.es/). ACM records wrist temperature (T), motor activity (A), body position (P), and environmental light exposure (L) rhythms during a whole week. Sleep was inferred from the integrated variable TAP (from temperature, activity, and position). Non-parametric analyses of TAP and estimated sleep yielded indexes of interdaily stability (IS), intradaily variability (IV), relative amplitude (RA), and a global circadian function index (CFI). Mid-sleep and mid-wake times were estimated from the central time of TAP-L5 (five consecutive hours of lowest values) and TAP-M10 (10 consecutive hours of maximum values), respectively. The most discriminative parameters, determined by ANOVA, Chi-squared, and information gain criteria analysis, were employed to build a decision tree, using machine learning. This model differentiated between healthy controls, DSPD and three insomnia subgroups (compatible with onset, maintenance and mild insomnia), with accuracy, sensitivity, and AUC >85%. In conclusion, circadian parameters can be reliably and objectively used to discriminate and characterize different sleep and circadian disorders, such as DSPD and OI, which are commonly confounded, and between different subtypes of PI. Our findings highlight the importance of considering circadian rhythm assessment in sleep medicine.

Urban environment influences on stress, autonomic reactivity and circadian rhythm: protocol for an ambulatory study of mental health and sleep

“…The cutting-edge research in this field integrates melatonin samples with multisensor ambulatory circadian monitoring and leverages machine learning models to predict the impact of light exposure on melatonin secretion ( 92 ). Similarly, machine learning models can also be used to detect circadian rhythms of sleep, motor, and autonomic functioning from multisensor circadian monitoring ( 93 ) to predict sleep and circadian disorders ( 69 ). Although these novels methods are under development, they have yet to be integrated in protocols investigating the effects of the built, socio-environmental, and home environments on mental and sleep health (see Figure 3 for a summary of new bio-signals to be considered in future studies).…”

“…Following previous protocols, to compensate for the lack of light spectrum detection with the chosen device, measurements over 1000 Lux will be considered as sun exposure whereas lower lux counts will be classified as artificial light ( 67 ). Following previous protocols, skin temperature and accelerometer, which are both input and output circadian signals, and light exposure, which is an input signal, will be used for continuous ambulatory circadian monitoring of participants ( 68 , 69 ). In the absence of hormonal samples to directly measure the effects of circadian alignment on the phases and amplitudes of the main circadian hormones (such as melatonin or cortisol), ambulatory circadian monitoring and sleep time patterns (onset and mid-sleep) are used instead as the main proxies for circadian entrainment (i.e., the synchronization of the internal biological clock to external time cues, such as the natural dark-light cycle).…”

A machine learning model for multi-night actigraphic detection of chronic insomnia: development and validation of a pre-screening tool

“…Differentiating sleep and wake with the absence of movement using a single wrist-worn actigraphy sensor is a long-standing problem. To do this may require multimodal systems that can monitor vital physiological signals in addition to motor activity [ 49 ]. However, in this study, we have restricted our approach to a single wrist-worn sensor to allow a less obtrusive long-term monitoring alternative for pre-screening individuals for insomnia.…”

Estimating circadian rhythm across infancy: Development of intra- and interdaily stability, and relative amplitude

“…IS, IV and RA are thus promising indicators for sleep-wake rhythm development. They have been used considerably to assess the status of the circadian rhythmicity in various populations, i.e., patients with Alzheimer's disease, diabetes, Down Syndrome, bipolar disorder, and also infants (Lovos et al, 2021;Rock et al, 2014;Van Someren et al, 1999;Witting et al, 1990;Zornoza-Moreno et al, 2011 and was implemented in different populations (Bandín et al, 2014;Rodriguez-Morilla et al, 2019;Zornoza-Moreno et al, 2011.…”

“…with non-parametrical variables, such as IS, IV, RA and CFI, which are particularly specific and useful to estimate circadian rhythm(Ortiz- Tudela et al, 2010;Rodriguez-Morilla et al, 2019;Zornoza-Moreno et al, 2011). IV significantly decreases with age in infants, while RA, IS, and CFI variables were shown to increase between 3 and 6 months old(Zornoza-Moreno et al, 2011).…”