From Pen Strokes to Sleep States: Detecting Low-Recovery Days Using Sigma-Lognormal Handwriting Features

A research team led by Koichi Kise from Osaka University, in collaboration with the University of Colorado Boulder, has published a groundbreaking study demonstrating that everyday handwriting contains measurable signals of physiological recovery. The team developed a personalized binary classification framework that uses features derived from the Sigma-Lognormal model—a mathematical representation of the neuromotor process behind pen strokes—to detect 'low-recovery' days. In their 28-day in-the-wild study involving 13 university students, participants recorded their handwriting three times daily using a digital device while simultaneously wearing an Oura Ring to track nocturnal cardiac indicators like Heart Rate Variability (HRV) and sleep duration.

The AI system was trained to identify the lowest quartile of recovery for four key metrics: HRV, lowest heart rate, average heart rate, and total sleep duration. Using Leave-One-Day-Out cross-validation, the model's performance, measured by PR-AUC (Precision-Recall Area Under the Curve), significantly exceeded the 0.25 random baseline for all four variables after False Discovery Rate correction. The strongest correlations were found with the cardiac-related variables (HRV and heart rate), not sleep duration. Crucially, the classification accuracy remained consistent across different writing tasks and times of day, indicating that the recovery signal is embedded in fundamental movement dynamics, not specific content.

This research, published on arXiv, opens a new frontier for passive, device-independent health monitoring. By analyzing something as simple and frequent as handwriting—a task performed billions of times daily—the system could provide continuous, unobtrusive insights into an individual's autonomic nervous system state and recovery readiness without requiring dedicated wearables.