Combination of Empirical Mode Decomposition and Hjorth Parameters for Prediction of Preterm Labor using Electrohysterogram Signals

Timely predicting preterm labor is vital for increasing the chance of neonatal survival rates and promoting maternal well-being. In this study, we explore the potential of Hjorth parameters for predicting preterm labor using the electrohysterogram (EHG) signals. Our proposed algorithm begins by decomposing the EHG signals into four modes through empirical mode decomposition. Subsequently, for each mode, we extract Hjorth parameters—activity, mobility, and complexity— and input them into a random forest classifier for discrimination between term and preterm labor. The comparative analysis of these three parameters highlights the superiority of complexity over activity and mobility, resulting in higher accuracy (85% compared to 73% and 77%). These findings underscore the effectiveness of Hjorth parameters in predicting preterm labor using EHG signals.