R functions used for data analysis for the article "Potential exoskeleton uses for reducing low back muscular activity during farm tasks" published in American Journal of Industrial Medicine, DOI: 10.1002/ajim.23180
Two primary steps for EMG are:
1. Signal processing for each participant
- For generating exposure summary metrics, emg.main.R processes time series data as collected in the field, uses parallel computing to process all tasks at once, its functions are:
- emg.filter.R filters signal with Butterworth bandpass
- summary.psd.R derrives frequency-domain parameters to help evaluate whether the signal has an error.
- signal.amplitude.R calculates root-mean-squares of the signal
- For visualization, emg.presentation.R creates plots for individual task-participant combinations
2. Statistical analysis for all participants
- Statistical analysis was conducted for laboratory-based simulated tasks and actual farm tasks separately.
- For simulated tasks, emg.sim.R, Wilcoxon signed-rank tests were used.
- For farm tasks, emg.farm.R, Shapiro-Wilk tests proves normality on log-transformed EMG measures, and linear regression was conducted with exoskeleton presence as predictor while age, sex, BMI and participant ID as counfounding factors.
- For visualization, bar charts and scatter plots were made with the following codes:
- data_summary.R was borrowed from external source to help create graphs.
- emg_visualize_manuscript.R was applied to make graphs for publication.