How Effective Are Forecasting Models in Predicting Effects of Exoskeletons on Fatigue Progression?

Area of Science:

• Biomechanics
• Ergonomics
• Data Science

Background:

• Forecasting models can predict physiological demands, aiding intervention development.
• Exoskeleton (EXO) technology shows promise in mitigating physical strain during occupational tasks.

Purpose of the Study:

• To implement and compare forecasting models (ARIMA, Facebook Prophet) for predicting fatigue progression during exoskeleton-assisted tasks.
• To assess the impact of exoskeleton assistance on delaying fatigue onset and progression.

Main Methods:

• Nine participants performed intermittent 45° trunk flexion tasks with and without exoskeleton assistance.
• Perceived fatigue and low-back muscle activity data were collected.
• Autoregressive Integrated Moving Average (ARIMA) and Facebook Prophet models were applied using univariate and multivariate data.

Main Results:

• Univariate Prophet models demonstrated superior performance in predicting fatigue levels (RMSE: 0.62-0.67).
• Exoskeleton assistance significantly reduced the slope of fatigue progression by approximately 48-52% over 20 trials.
• Forecasting models indicated median fatigue reduction benefits of 43-54% with exoskeleton assistance.

Conclusions:

• Forecasting models, particularly Facebook Prophet, can effectively predict fatigue progression in occupational settings.
• Exoskeleton assistance substantially delays fatigue, highlighting its potential for enhancing worker health and preventing injuries.
• This research suggests practical applications for forecasting in workforce health monitoring and intervention assessment.