Introduction
In the field of sleep research, the Sleep Regularity Index (SRI) has emerged as a vital metric for assessing health risks, particularly in older populations. However, a technical correction has been issued for the original 2018 study published in Scientific Reports. This amendment addresses a mathematical error in the formula used to calculate the "sleep midpoint," a critical component of sleep timing analysis.
The Challenge of Circular Statistics
Analyzing sleep timing presents a unique challenge: time is a periodic variable. Standard arithmetic means fail when data points cross the midnight threshold. For instance, the average of 11:00 PM (23:00) and 1:00 AM (01:00) should be Midnight (00:00). A linear calculation, however, would incorrectly yield 12:00 PM (Noon).
To resolve this, researchers employ circular statistics, treating time as points on a circle. This requires trigonometric functions to convert time into coordinates $(x, y)$ on a unit circle, calculating the mean angle, and then converting that angle back into a 24-hour format.
Correcting Equation 3
The original publication contained a typographical error in Equation 3, which defines the sleep midpoint. The formula incorrectly utilized a plus sign ($+$) within the arctan2 function, effectively merging two distinct coordinate summations into a single value. This rendered the formula mathematically invalid for its intended purpose.
The corrected formula replaces the plus sign with a comma ($,$), properly defining the two arguments required for the arctan2 function:
$$ \text{Sleep Midpoint} = \frac{1440}{2\pi} \text{arctan2} \left( \sum \sum s_{i,j} \sin \frac{2\pi t_i}{1440} , \sum \sum s_{i,j} \cos \frac{2\pi t_i}{1440} \right) $$
Technical Implications
The arctan2(y, x) function is a standard computational tool that determines the angle of a point $(x, y)$ across all four quadrants ($-\pi$ to $\pi$). By providing the sine-summation and cosine-summation as separate arguments, the formula correctly identifies the mean direction of the sleep data points.
Furthermore, the authors have added an explicit definition for arctan2(x, y) to clarify that it denotes the angle between the coordinates and the positive x-axis. This ensures that the resulting midpoint can be accurately mapped back to the 1,440 minutes in a day.
Conclusion
While the core findings of the original study regarding the SRI and cardiometabolic risk remain robust, this correction is essential for researchers seeking to replicate the methodology or apply the SRI in new contexts. It ensures that the mathematical foundation for calculating sleep timing is precise and consistent with the principles of circular statistics.
Summary updated