Software System for Prediction of Shiftwork Alertness

Heitmann, Anneke

Abstract

A software system which Circadian Technologies has patented will be developed into a practical tool for accurately predicting alertness levels at work considering circadian and homeostatic aspects of sleep and alertness. The core algorithms of the software system are based on published data from human sleep laboratory studies. These algorithms will be refined to accommodate irregular sleep-wake-work pattern using a large training data set from 50 railroad engineers with irregular work pattern drawn from Circadian Technologies' extensive data base on sleep-wake-work patterns and alertness levels at work. Interfaces will be developed to allow for efficient input processing and presentation of multiple days of individual data. The predictive capability of the software system will be evaluated based on comparisons with other alertness models,, and cross validation will be performed by applying the developed algorithms to four test data sets from several additional railroads. This software system will be designed to be used as a tool to help minimize sleepiness at work (for example by designing more bio-compatible work schedules or by timing the effective use of fatigue countermeasure such as preventive naps) and can potentially benefit health and safety of a large portion of working society.

Proposed Commercial Applications

The progressive transition into a 24-hour society creates a large market for a software system for predicting alertness at work. This software system would be an attractive tool in shiftwork operations for mangers who aim to design bio-compatible work schedules in order to increase health and safety. It can contribute to public education regarding chronobiological and homeostatic aspects of sleep and sleepiness, and it can help increase awareness about the practical implications of sleepiness at work.