Biosensor-Assisted Method for Abdominal Syndrome Classification Using Machine Learning Algorithm.

Area of Science:

• Gastroenterology and Biomedical Engineering
• Physiology and Medical Signal Processing

Background:

• Digestive system disorders like dyspepsia and nausea are prevalent globally.
• Accurate diagnosis of gastric anomalies often requires invasive procedures.
• Electrogastrogram (EGG) measures stomach muscle electrical activity, offering a non-invasive diagnostic approach.

Purpose of the Study:

• To develop and validate a novel computational method for analyzing electrogastrogram (EGG) signals.
• To differentiate between normal and abnormal gastric electrical rhythms indicative of digestive disorders.
• To establish EGG analysis as a reliable tool for diagnosing stomach conditions.

Main Methods:

• Collected EGG data from healthy individuals and patients with various digestive conditions (dyspepsia, nausea, ulcer, etc.).
• Utilized Continuous Wavelet Transform (CWT) with a genetic algorithm (db4) for EGG signal pattern extraction in MATLAB.
• Employed an Adaptive Resonance Classifier Network (ARCN) to classify EGG signals based on alertness parameter (μ).

Main Results:

• The CWT analysis successfully generated 3D plots visualizing EGG signal cycles, with peaks indicating signal presence.
• The ARCN achieved a high diagnostic accuracy of 95.45%, with sensitivity at 92.45% and specificity at 87.12%.
• The study demonstrated the capability to categorize EGG signals and identify abnormal patterns.

Conclusions:

• The proposed EGG analysis method provides an accurate and non-invasive means for diagnosing digestive system problems.
• This computational approach can serve as a valuable medical tool, potentially reducing the need for invasive diagnostic treatments.
• The high accuracy and sensitivity/specificity suggest clinical utility in gastroenterology.