Enhancement of Stress Analysis Performance using Respiration Information

Aims: The principal purpose of this study is to identify stress through the analysis of both electrocardiogram (ECG) and respiration signals (RESP). Various parameters were extracted and stress detection is performed using popular data analysis algorithms like Support Vector Machine (SVM) and Random Forest (RF) classifiers.Methods: Private Dataset of ECG and RESP signal is developed using BIOPAC MP160 data acquisition system. Both female and male subjects were included in experiment. All participants were instructed to undergo though four different conditions like resting, reading an article silently, listening Indian Classical Instrumental music and arithmetic calculation. R-R peak interval from ECG and peak-peak interval from RESP signal were used to compute basic linear parameters in time domain and frequency domain. Next, two-machine learning classifiers were applied to detect different stress leading activities. Conventional model performance analysis parameters are calculated and presented in result.Results: Data was collected from 14 participants under different circumstances. The findings indicated that the best outcomes were achieved incorporating a variety of features extracted from both ECG and RESP signals. These results showed the highest levels of accuracy (95 ± 6.12), precision (94.28 ± 6.99), recall (100 ± 0.0), and F1-score (96.92 ± 3.76) using ECG and RESP signals.Conclusion: In summary, the existing evidence on the cardio-respiratory system indicates that stress affects both ECG (electrocardiogram) and respiration. This underscores their potential utility in objectively evaluating psychological well-being and stress levels.