Computer methods and programs in biomedicine · 2026
LIBPhys-EGG: An open-access multichannel electrogastrography and accelerometry dataset for reproducible research in gastrointestinal electrophysiology.
СтатьяНазвание и аннотация: Перевод пока не подготовлен
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Braga Rodrigo, Sousa Sara, Ao Nianfei, Osório Daniel, Xu Jun, Liu Hui, Gamboa Hugo
Аннотация
BACKGROUND AND OBJECTIVE: Surface electrogastrography records gastrointestinal myoelectric activity non-invasively from the abdominal surface. Despite decades of investigation, clinical adoption remains limited by low signal-to-noise ratio, motion artifacts, and the lack of standardized public datasets. This work presents a multichannel electrogastrography dataset acquired using a wearable sensor belt, synchronized with tri-axial accelerometry, designed to support reproducible research in gastrointestinal signal processing and machine learning.
METHODS: The dataset (LIBPhys-EGG) includes 30 recording sessions from 15 healthy adults, totaling 111.8 h of multichannel physiological data. Primary session duration averaged 5.16 ± 1.22 h, covering fasting, meal ingestion, and post-prandial periods. Four bipolar electrogastrography channels and a tri-axial accelerometer were sampled at 1000 Hz. We release raw data with synchronized annotations, a recommended pipeline, feature selection methodology, and a benchmark classification task distinguishing pre- versus post-prandial states.
RESULTS: LIBPhys-EGG captured strong physiological changes post-meal. Features measuring signal power and amplitude variability increased substantially across 13 high-quality recordings from 7 unique participants, with large effect sizes (Cohen's dz up to 2.06, p<0.002). Random Forest achieved 0.93 accuracy and 0.966 AUC-ROC using 14 features on recording-level split data; for the same model, leave-one-subject-out validation yielded 0.690 ± 0.198 accuracy.
CONCLUSIONS: This open dataset offers high-resolution, multichannel sEGG with synchronized motion signals and event annotations, enabling development and benchmarking of noise-robust signal processing and machine-learning methods. It also facilitates analysis of spatial signal distribution and electrode placement. We anticipate this resource will accelerate non-invasive GI research and contribute to standardized protocols for future studies.
METHODS: The dataset (LIBPhys-EGG) includes 30 recording sessions from 15 healthy adults, totaling 111.8 h of multichannel physiological data. Primary session duration averaged 5.16 ± 1.22 h, covering fasting, meal ingestion, and post-prandial periods. Four bipolar electrogastrography channels and a tri-axial accelerometer were sampled at 1000 Hz. We release raw data with synchronized annotations, a recommended pipeline, feature selection methodology, and a benchmark classification task distinguishing pre- versus post-prandial states.
RESULTS: LIBPhys-EGG captured strong physiological changes post-meal. Features measuring signal power and amplitude variability increased substantially across 13 high-quality recordings from 7 unique participants, with large effect sizes (Cohen's dz up to 2.06, p<0.002). Random Forest achieved 0.93 accuracy and 0.966 AUC-ROC using 14 features on recording-level split data; for the same model, leave-one-subject-out validation yielded 0.690 ± 0.198 accuracy.
CONCLUSIONS: This open dataset offers high-resolution, multichannel sEGG with synchronized motion signals and event annotations, enabling development and benchmarking of noise-robust signal processing and machine-learning methods. It also facilitates analysis of spatial signal distribution and electrode placement. We anticipate this resource will accelerate non-invasive GI research and contribute to standardized protocols for future studies.