Generalised linear model of periodic limb movements

During sleep, the lower extremities exhibit periodic repetitive movements which are referred to as Period Limb Movement (PLM). Polysomnography (PSG) is the gold standard for diagnosing periodic limb movement disorder. The frequency of PLM episodes per hour of sleep (PLMI) determines the severity of the condition. PLM are generated by a dynamic process, however PLMI measures only the average PLM rate and does not capture the dynamic properties of PLM. Here, we characterise PLM dynamics using a generalised dynamic model as a function of sleep stage, timing of past PLM events and adjacent sleep disordered-breathing events. We analysed PSG recordings of 237 men and 222 women enrolled in the Multi-ethnic Study of Atherosclerosis (MESA) dataset to model dynamic PLM features. We statistically analysed whether these dynamics are associated with sex, age, and BMI. Modelling suggests instantaneous PLM rates are greater in men than women and higher in N1 and N2 non-rapid eye movement sleep than N3 and rapid eye movement sleep. The generalised model constitutes statistically robust approach towards the characterisation of periodic limb movement.Clinical Relevance - The generalised model may enable differentiated diagnostics of periodic limb movement disorder.