Automatic Classification of Normal and Dysphagic Swallows Based on the sEMG Signals

Swallow events classification is significant to improve the performance of swallow detection in digital health technologies. Our work consists of two major contributions: First, we perform data preprocessing techniques and transform the raw sEMG signals into feature vectors before using them to train an ML classifier. It is important for developing a data exploration project since preprocessing the data can often help with getting rid of the noise. Besides, properly optimized feature extraction is the key to effective model construction in machine learning. Second, we develop an ML swallow-detection algorithm to classify swallow events that can use algorithms like Artificial Neural Network (ANN), fuzzy classifier, Support Vector Machines (SVM), etc. Improving the accuracy of swallowing detection will lead to more robust digital health software systems to support patients who are experiencing swallowing disorders because of conditions like head and neck cancer, stroke or Parkinson’s Disease.

Faculty Supervisor:

Linglong Kong

Student:

Partner:

True Angle Medical Technologies, Inc.

Discipline:

Mathematics

Sector:

Health and Related Sciences & Technology; Manufacturing

University:

University of Alberta

Program: