Hidden Markov models for monitoring circadian rhythmicity in telemetric activity data

Abstract

Wearable computing devices allow collection of densely sampled real-time information on movement enabling researchers and medical experts to obtain objective and non-obtrusive records of actual activity of a subject in the real world over many days. Our interest here is motivated by the use of activity data for evaluating and monitoring the circadian rhythmicity of subjects for research in chronobiology and chronotherapeutic healthcare. In order to translate the information from such high-volume data arising we propose the use of a Markov modelling approach which (i) naturally captures the notable square wave form observed in activity data along with heterogeneous ultradian variances over the circadian cycle of human activity, (ii) thresholds activity into different states in a probabilistic way while respecting time dependence and (iii) gives rise to circadian rhythm parameter estimates, based on probabilities of transitions between rest and activity, that are interpretable and of interest to circadian research.

Tele-Monitoring of Cancer Patients’ Rhythms during Daily Life Identifies Actionable Determinants of Circadian and Sleep Disruption

“…A hidden Markov model was fitted to 5-min aggregated chest rest-activity data, which retrospectively infers the times an individual spent in three different states, that we will define as an inactive/rest (IA), moderately active (MA), and highly active (HA) state, and are specific to the person [ 25 ]. The HMM further produces a variety of numerical quantifiers that are of interest to circadian rhythm in rest-activity, such as: The mid values of the MA and HA states which indicate daily activity levels; The rhythm index (RI), with values ranging between 1, corresponding to best average quality and regularity of the IA state, and 0, corresponding to poor quality and absence of a consistent rest state in the pattern; The average center-of-rest time of the IA state; P1-1, the estimated probability of staying in the IA state (state 1); when having previously been in this state.…”

“…The I < O value of cancer patients was first computed using the first 72 h of recordings from the chest sensor ((I < O) 72h ), so as to categorize them in the (I < O) high group if (I < O) 72h was above 97.5% or in the (I < O) low group if (I < O) 72h was below or equal to 97.5%, which led to referral to sleep and physiotherapy clinics. I < O was also computed at the end of the study over the 7 days of chest activity recordings for all 58 participants A hidden Markov model was fitted to 5-min aggregated chest rest-activity data, which retrospectively infers the times an individual spent in three different states, that we will define as an inactive/rest (IA), moderately active (MA), and highly active (HA) state, and are specific to the person [25]. The HMM further produces a variety of numerical quantifiers that are of interest to circadian rhythm in rest-activity, such as:…”

“…Such a domomedicine platform proved to be acceptable and technically reliable in field studies involving 223 people, including shift workers and cancer patients [23]. Appropriate statistical methods were developed for estimating CTS function in an individual subject during their daily life [24,25]. Here, we identify the main actionable determinants of circadian and sleep disruption through tracking two circadian biomarkers in real time in two clinical studies involving a pooled sample of size 58 comprising both cancerous and non-cancerous individuals.…”

Influence of sleep-wake cycle on body mass index in female shift-working nurses with sleep quality as mediating variable

“…This means that the activity level during sleep periods is lower than the median percentage of that during awake periods; 100% indicates that the sleep period is uninterrupted, and lower I<O values mean greater disruptions in the sleep-wake cycle. In this way, we could objectively measure the interdaily stability of sleep and activity 21,22) .…”

Wearable Device Heart Rate and Activity Data in an Unsupervised Approach to Personalized Sleep Monitoring: Algorithm Validation

“…Temporal dependency can be learned, and the parameters of the model are interpretable. In the context of bioinformatics, HMMs have been applied to monitor circadian rhythmicity using physical activity data to characterize interindividual variability [ 28 ]. In other high-frequency physiological data collected during PSG—such as electroencephalography, electrooculography, and electromyography—HMMs have been used to classify sleep stages [ 29 , 30 ].…”