Temperature- Induced Surface Electromyography Changes: A Study on Physiological Responses and Neural Network Predictions

In this study, we explore the correlation between variations in temperature within a controlled thermal chamber and the resulting surface electromyography (sEMG) signals. Participants introduced their arms into a chamber for sEMG signal recording. These signals underwent meticulous pre-processing to eliminate direct current (DC) components and mitigate electromagnetic interference while also extracting the root mean square (RMS) envelope. We used the Approximated Generalized Likelihood Ratio (AGLR) algorithm to quantify the likelihood that observed sEMG signal changes were due to temperature fluctuations. Towards modeling the correlation, the NARX neural network structure is used to anticipate the variations in sEMG signals induced by alterations in the temperature of the surrounding environment. This model was meticulously trained on the refined sEMG data as well as corresponding temperature measurements. Initial results suggest that the NARX neural network model is highly promising in deciphering the complex physiological responses triggered by changes in environmental conditions. This research not only establishes the fundamental groundwork for further investigation into the complex interplay between human physiological reactions and external temperature stimuli but also holds significant implications for fields spanning ergonomics, rehabilitation, and beyond.