COVID-19 Prediction With Machine Learning Technique From Extracted Features of Photoplethysmogram Morphology
Nazrul Anuar Nayan1,2, Choon Jie Yi1, Mohd Zubir Suboh1
1Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia, Bangi, Malaysia.
Frontiers in Public Health
|August 5, 2022
Summary
This study used photoplethysmogram (PPG) signals from wearable devices to detect COVID-19. Machine learning models, particularly artificial neural networks, achieved high accuracy in identifying infected individuals based on PPG features.
Area of Science:
- Biomedical Engineering
- Computational Biology
- Infectious Disease Research
Background:
- COVID-19 poses significant global health risks, including respiratory complications.
- Wearable devices offer potential for remote monitoring and early detection of infectious diseases like COVID-19.
Purpose of the Study:
- To investigate the correlation of photoplethysmogram (PPG) morphology in COVID-19 patients versus healthy individuals.
- To develop a machine learning model for COVID-19 prediction using PPG data.
Main Methods:
- Collected PPG data from 86 subjects (43 COVID-19 cases, 43 controls).
- Extracted and analyzed 20 morphological features from PPG signals.
- Utilized machine learning algorithms including discriminant analysis, k-nearest neighbor, decision tree, support vector machine, and artificial neural network (ANN).
Main Results:
- Significant differences in PPG morphology were observed between COVID-19 cases and controls, with 12 out of 20 features showing statistical significance.
- The artificial neural network (ANN) model demonstrated superior performance, achieving 95.45% accuracy, 100% sensitivity, and 90.91% specificity.
- Key distinguishing features included dicrotic-systolic time interval, onset-dicrotic amplitude, and systolic-onset time interval.
Conclusions:
- PPG signal morphology contains discriminative features for COVID-19 detection.
- A low-cost pulse oximeter combined with machine learning, specifically ANN, can form an effective COVID-19 prediction model.
- Wearable PPG technology shows promise for non-invasive, remote COVID-19 screening and monitoring.
