dlmoR: An Open-Source R Package for the Dim-Light Melatonin Onset (DLMO) Hockey-Stick Method

Area of Science:

• Chronobiology and Sleep Research
• Computational Biology and Bioinformatics
• Biostatistics and Data Analysis

Background:

• The dim-light melatonin onset (DLMO) is a critical biomarker for human circadian timing.
• Existing methods for DLMO calculation, such as the hockey-stick method, have limitations.
• There is a need for robust, open-source tools for accurate DLMO estimation.

Purpose of the Study:

• To introduce dlmoR, an open-source R implementation of the hockey-stick method for DLMO calculation.
• To validate dlmoR against existing methods and assess its performance on diverse datasets.
• To explore the sensitivity of the hockey-stick algorithm to various data and parameter variations.

Main Methods:

• Developed dlmoR as a clean-room implementation of the hockey-stick algorithm in R.
• Validated dlmoR against a reference binary executable using 112 melatonin time series datasets.
• Performed sensitivity analyses by systematically altering sampling schedules, thresholds, data completeness, and noise.

Main Results:

• dlmoR demonstrated high agreement with the reference implementation (circular correlations > 0.96).
• Paired t-tests showed no significant systematic differences or bias between dlmoR and the reference method.
• Sensitivity analyses revealed that moderate data perturbations yielded stable DLMO estimates, while significant alterations led to substantial delays or failed estimations.

Conclusions:

• dlmoR provides a reliable, reproducible, and enhanced tool for DLMO estimation.
• The hockey-stick algorithm exhibits fundamental limitations in handling data variations and identifying the melatonin rise.
• The findings highlight the necessity for developing novel, uncertainty-aware approaches for DLMO estimation.