Drowsiness Detection for Office-based Workload with Mouse and Keyboard Data

TL;DR

This paper introduces a non-invasive, office-friendly method for detecting drowsiness by analyzing keyboard, mouse, and eye movements with machine learning, correlating biometric data with self-reported sleepiness levels.

Contribution

It proposes a novel approach combining biometric features from common office devices and eye tracking to assess drowsiness without intrusive sensors.

Findings

Strong correlation between biometric predictions and self-reported drowsiness.

Feasibility of real-time drowsiness monitoring in office environments.

Potential for non-intrusive fatigue detection using standard office equipment.

Abstract

Non-invasive devices involved in the detection of drowsiness generally include infrared camera and Electroencephalography (EEG), of which sometimes are constrained in an actual real-life scenario deployments and implementations such as in the working office environment. This study proposes a combination using the biometric features of keyboard and mouse movements and eye tracking during an office-based tasks to detect and evaluate drowsiness according to the self-report Karolinska sleepiness scale (KSS) questionnaire. Using machine learning models, the results demonstrate a correlation between the predicted KSS from the biometrics and the actual KSS from the user input, indicating the feasibility of evaluating the office workers' drowsiness level of the proposed approach.