Data from: Marked Point Process Secretory Events Statistically Characterize Leptin Pulsatile Dynamics

We utilized leptin data collected by Kok et al. in their previous clinical studies. The participants were 18 healthy premenopausal women with obesity (BMI 30.1 − 40.5kg/m2, age 22–51 years with mean = 37.5 ± 1.7) who were not under the influence of any medication or drugs. Subjects with acute or chronic disease, depression, head trauma, habits of smoking or alcohol consumption, recent transmeridian flights, nightshift work,weight change, blood donation, or participation in another clinical trial were excluded. To minimize potential confounding factors, the data were gathered during the early follicular phase of their menstrual cycles, and all subjects maintained a eucaloric diet and adhered to the same sleep schedule, from 11:00 PM to 7:30 AM. The study period spanned 24 hours, from 9 AM to 9 AM on the next day, following 7 days of placebo treatment. Hormone concentration levels were measured from blood samples using radioimmunoassay, which has a detection limit of 0.5 ng/L, at 10-minute intervals throughout this period. Four weeks later, the same protocol was repeated with the same subjects. However, this time, each participant received a 2.5 mg dose of bromocriptine twice daily for 7 days to assess the effects of bromocriptine. In their previous study, Reddy et al. deconvolved these plasma leptin measurements to understand the underlying secretion of leptin. Specifically, they extracted the timings and amplitudes of leptin secretory pulses throughout the duration of the recording while considering physiological constraints of leptin secretion such as its pulsatile and sparse nature. There are normally 20 to 50 secretory pulses in 24 hours, and the pulses are discrete. They modeled the leptin secretion process as a state-space model, the details of which are presented in later sections. To perform the deconvolution and recover the leptin pulses, the parameters of the model are solved using the FOCUSS+ algorithm and Generalized Cross Validation. For the purpose of this study, we applied our statistical models to these extracted pulses for each of the 18 subjects.