SOM Clustering and Modelling of Australian Railway Drivers’ Sleep, Wake, Duty Profiles

摘要: Two SOM ANN approaches were used in a study of Australian railway drivers (RDs) to classify RDs’ sleep/wake states and their sleep duration time series profiles over 14 days follow-up. The first approach was feature-based that clustered the most frequently occurring patterns sleep. second created RD networks sleep/wake/duty/break feature parameter vectors between-states transition probabilities via multivariate extension mixture distribution (MTD) model, accommodating covariate interactions. SOM/ANN found 4 clusters RDs whose differed significantly. Generalised Additive Models for Location, Scale Shape 2 outcomes confirmed break onset times, hours next duty are significant effects which operate differentially across groups. Generally increases between 10 am pm, when since exceeds 1 day. These factors determining current sleep, have differential impacts clusters. Some groups catch up after night shift, while others do so before shift. Sleep is governed by RD’s anticipatory behaviour scheduled onset, driver experience, age domestic scenario. This has clear health safety implications rail industry.