Ever wondered how to deal with darkness?
A new publication from our Dr Johannes Zauner addresses a critical methodological challenge in circadian and myopia research: how to properly handle zero-lux measurements that routinely occur in personal light exposure studies.
Zero-lux readings matter. In free-living light measurements, true zeros routinely occur and can't simply be dropped or ignored when applying log-based analyses. When people sleep, move through windowless spaces, or spend time under covers, these darkness periods contain important information about light exposure patterns.
Our research compared different approaches to visualizing and modelling this zero-inflated light data. We tested four visualization scales and three modelling strategies to find the best ways to handle these challenging but crucial measurements.
Key findings:
Symlog transformations are best for visualization, showing zero-lux values while preserving the full dynamic range from very low to very high light exposures
Adding –1 log₁₀ lux to zero readings allows robust statistical modelling on logarithmic scales
Tweedie models offer a unified solution that handles zeros and skew without requiring data transformation
The methods extend beyond light exposure research to other biological rhythm data that exhibit zero inflation, such as accelerometery in sleep studies or activity monitoring in animal models.
LightLogR: your toolbox for zero-inflated light data. Our open-source R package LightLogR implements these visualization and modelling solutions, ready for circadian, sleep, and myopia researchers to deploy.
Proper treatment of darkness and low-light exposures is critical for understanding human circadian entrainment, light-related health outcomes, and myopia development under real-world conditions.
https://journals.sagepub.com/doi/10.1177/07487304251336624