A leptin-based Bayesian inference of a pro-satiety state reflects a basal circadian rhythm in women with obesity

Leptin, primarily secreted by adipose tissue, is a critical hormone involved in regulating energy balance and food intake by inducing satiety. Although several hormones influence satiety, leptin plays a dominant role in long-term satiety regulation.

We apply a state-space estimation framework using Bayesian filtering to infer continuous, long-term pro-satiety states from plasma leptin concentrations collected from premenopausal women with obesity. Our approach adopts methodologies previously applied to biosignals such as skin conductance and cortisol data to estimate latent states, leveraging the features in the leptin secretory pulses and plasma leptin levels. Additionally, we investigate the potential influence of meals, sleep, and bromocriptine treatment on the pro-satiety states. We introduce the High Satiety Index (HSI), a direct, long-term satiety measure based on leptin secretion dynamics, minimizing biases inherent in conventional assessment methods.

Comparisons of the estimated state in different time windows show that the pro-satiety state inferred by leptin secretion is significantly higher during sleep, aligning with a circadian rhythm. The estimated state does not show a significant variation in response to meal intake or bromocriptine treatment.

The leptin-based estimator reflects basal variations of satiety in women with obesity. This framework shows the feasibility of applying Bayesian filtering to track satiety and can be further developed to perform a multimodal estimation.