Endocrine pulse identification using penalized methods and a minimum set of assumptions

Area of Science:

• Endocrinology
• Computational Biology
• Biostatistics

Background:

• Accurate detection of hormone secretion episodes is crucial for understanding endocrine system function and dysfunction.
• Identifying hormone pulses from short-interval sampling data presents significant analytical challenges.
• Existing models require biologically plausible assumptions to yield reliable hormone secretion and clearance results.

Purpose of the Study:

• To develop and validate a novel computational method for identifying hormone secretion events.
• To improve the accuracy and resolution of hormone pulse detection, especially during periods of low activity.
• To compare the performance of the new method against a widely used existing model.

Main Methods:

• Utilized a penalized nonlinear least-squares approach to determine the number of hormone secretion events.
• Assumed a sparse secretion pattern without predefining pulse shape or frequency.
• Applied the VisPulse method to luteinizing hormone (LH), cortisol, growth hormone, and testosterone data.

Main Results:

• The VisPulse method demonstrated strong performance across multiple hormones, including LH, cortisol, growth hormone, and testosterone.
• Analysis of LH data showed high correlation between modeled and measured concentrations, with sparse secretion patterns and minimal residuals.
• Benchmarking against AutoDecon revealed VisPulse's superior accuracy, particularly in detecting silent periods and small secretion events, evidenced by higher sensitivity and selectivity metrics.

Conclusions:

• The VisPulse method provides an accurate and robust approach for detecting hormone secretion episodes.
• This novel method offers improved resolution for identifying subtle endocrine events compared to existing techniques.
• The findings support the utility of VisPulse for advancing the study of endocrine physiology and pathophysiology.