Using HMMS to Identify Groups in a Patient Population: A Simulation

This study assessed the reliability, through a simulation study, of using hidden Markov models (HMMs) to identify groups of chronic lower back pain (CLBP) subjects. Two HMMs were designed to describe the lifting patterns of CLBP subjects and pain-free controls during a repetitive lifting task. The simulation study was conducted to determine how reliably the HMMs can detect intentionally misclassified simulated lifting sequences. The results of the simulation studies indicate that the HMMs can reliably identify sequences to the correct model and that HMM classification procedure can be used on clinical time series data to identify groups within a population.