Hidden Markov Models and exponential smoothing forecasting showed promising results in simulations for predicting future stress episodes based on physiological signals.
Can Hidden Markov Models and exponential smoothing accurately forecast future stress episodes using physiological signals?
Hidden Markov Models and exponential smoothing show potential for forecasting stress episodes based on physiological signals from wearable sensors.
Work-related stress is normal and at low levels it can actually increase productivity. However, the accumulation of stress may have significant long-term behavioral and physical health consequences such as sleep deprivation, and anxiety disorder. According to the American Psychological Association, around 49% of the U.S. population suffers from chronic daily stress. In this paper, we explore the use of Hidden Markov Models (HMM), and exponential smoothing forecasting for the prediction of future stress episodes. Existing research on wearable sensors, data processing, and real time stress inferencing allow for collection of a ‘stress history’ time series which is used to train the methods presented here. A brief overview of methods is given, and prediction results are presented. Extensive simulations show promising results on stress forecasting, and potential future work on preventive stress interventions.
Jaimes et al. (Sun,) conducted a other in Stress. Hidden Markov Models (HMM) and exponential smoothing forecasting was evaluated on Prediction of future stress episodes. Hidden Markov Models and exponential smoothing forecasting showed promising results in simulations for predicting future stress episodes based on physiological signals.