Non-Invasive Motion Detection and Classification in NICU Patients using Ballistographic Signals from a Pressure Sensitive Mat

It appears your Web browser is not configured to display PDF files. Download adobe Acrobat or click here to download the PDF file.

Click here to download the PDF file.

Creator: 

Aziz, Samreen Umaiman

Date: 

2022

Abstract: 

Patient movements can cause motion artifacts on physiological signals and can result in false alarms in a continuous patient monitoring environment. This thesis explores the use of centre of pressure (COP) signals from a pressure sensitive mat to (a) detect patient movement in real-time, and (b) classify the upper/lower directionality of movement. For (a), the sum-distance travelled by the COP is tracked over time using a sliding window with data from seven patients. Window-boundary-suppression led to improved motion detection with precision = 0.84 and recall = 0.71 with a window of 10 seconds. For (b), seven features were derived from the COP, and feature selection was done using out-of-bag-error ranking and sequential forward selection. It was found that using a sample imputation approach of adding ~13 minutes of hand-annotated new subject data to the training set makes the classifier most useful, producing accuracy = ~87.29%, precision = 0.90, and recall = 0.84.

Subject: 

Engineering - Biomedical

Language: 

English

Publisher: 

Carleton University

Thesis Degree Name: 

Master of Applied Science: 
M.App.Sc.

Thesis Degree Level: 

Master's

Thesis Degree Discipline: 

Engineering, Biomedical

Parent Collection: 

Theses and Dissertations

Items in CURVE are protected by copyright, with all rights reserved, unless otherwise indicated. They are made available with permission from the author(s).