Modeling diurnal hormone profiles by hierarchical state space models

Area of Science:

• Endocrinology
• Chronobiology
• Biostatistics

Background:

• Adrenocorticotropic hormone (ACTH) exhibits complex diurnal patterns with both smooth circadian rhythms and pulsatile activities.
• Comparing these patterns between patient groups and healthy controls presents a statistical challenge.

Purpose of the Study:

• To develop and apply a hierarchical state space model for analyzing ACTH diurnal patterns.
• To statistically compare smooth circadian rhythms and pulsatile activities of ACTH between chronic fatigue syndrome/fibromyalgia patients and healthy controls.

Main Methods:

• A hierarchical state space model incorporating periodic smoothing splines for circadian rhythms and autoregressive processes for pulsatile activities.
• Functional random effects were used to accommodate a matched pair design.
• Kalman filter algorithms were employed for efficient parameter estimation and signal extraction via posterior means.

Main Results:

• The proposed model successfully extracted smooth circadian rhythms and pulsatile activities from noisy ACTH observations.
• No significant differences were found in the smooth circadian rhythms of ACTH between patients and controls.
• Patients with chronic fatigue syndrome and fibromyalgia demonstrated weaker pulsatile ACTH activities compared to healthy individuals.

Conclusions:

• The developed statistical model effectively differentiates between smooth and pulsatile components of ACTH diurnal patterns.
• Altered pulsatile ACTH activity, not circadian rhythmicity, may be a distinguishing feature in chronic fatigue syndrome and fibromyalgia.