A new fully automated and data driven pipeline for electrogastrogram (EGG) data analysis

Electrogastrography (EGG) is a technique for recording electrical activity of the stomach using surface electrodes [2]. In recent years, there has been growing interest in integrating stomach physiology, as measured via EGG, into psychological sciences [2]. However, EGG signals have historically been challenging to analyze due to the weak electrical activity of the stomach at the body surface and the presence of numerous motion and electronic artifacts [2]. We present a new, rigorous, and data-driven pipeline for analyzing EGG data. This pipeline is based on a recent time-frequency representation (TFR) technique called the Superlet transform [1], which provides significantly improved time and frequency resolution compared to traditional TFR methods such as wavelet transform or Short-Time Fourier Transform (STFT). Leveraging the enhanced time resolution of this technique and the fact that motion artifacts are highly energetic and localized in time, the pipeline can reliably identify artifact-contaminated recordings using an entropy measure (Tsallis entropy [3]) of the signal’s instantaneous energy as a frequency marginal of the Superlet TFR. Once artifacted segments are detected, they are removed through a combination of local Tukey fence analysis and a robust random cut forest algorithm [4]. After artifact removal, both classical EGG metrics—such as dominant frequency and power—and novel features exploring both frequency and time dimensions are extracted to construct a characteristic fingerprint for each recording channel. Additionally, time series features such as dominant frequency are tested against noise using surrogate Monte Carlo statistical testing. The entire procedure has been tested on EGG recordings with controlled motion artifacts and is fully automated, eliminating the need for manual artifact rejection—one of the most complex challenges in EGG data analysis. This approach enables a deeper and more robust characterization of stomach activity, potentially leading to new insights into the intricate connections between the gut and the brain.