Towards Multimodal Equipment to Help in the Diagnosis of COVID-19 Using Machine Learning Algorithms

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Infectious Disease Diagnostics

Background:

• COVID-19 diagnosis relies on RT-PCR, which is costly, time-consuming, and can overwhelm labs during high demand.
• Biomedical data (SpO2, temperature, heart rate, cough) coupled with machine learning show promise for inferring COVID-19 infection.
• There is a need for accessible and rapid screening tools to manage the pandemic.

Purpose of the Study:

• To introduce the Integrated Portable Medical Assistant (IPMA), a multimodal device for collecting biomedical data and cough sounds.
• To develop and evaluate machine learning algorithms for inferring COVID-19 diagnosis using data from the IPMA.
• To assess the usability and potential of the IPMA for broader respiratory disease screening.

Main Methods:

• The IPMA device was developed to collect oxygen saturation (SpO2), body temperature, heart rate, and cough sounds.
• Quadratic kernel-free non-linear Support Vector Machine (QSVM) and Decision Tree (DT) algorithms were applied to analyze the collected data.
• Three datasets were utilized, including data from the IPMA, and evaluated using Accuracy (ACC) and Area Under the Curve (AUC) metrics.

Main Results:

• QSVM and DT algorithms achieved up to 88.0% ACC and 0.85 AUC on existing datasets, and up to 99% ACC and 0.94 AUC for COVID-19 inference.
• When applied to IPMA-acquired data, these algorithms demonstrated 100% accuracy in inferring COVID-19 infection.
• Volunteer feedback indicated high usability for the IPMA, with System Usability Scale (SUS) and Post Study System Usability Questionnaire (PSSUQ) scores of 85.5 and 1.41, respectively.

Conclusions:

• The IPMA is a user-friendly, multimodal device capable of collecting vital biomedical signals and cough sounds for disease inference.
• Machine learning algorithms applied to IPMA data show exceptional accuracy (100%) for COVID-19 screening.
• The IPMA holds significant potential as a smart screening tool for COVID-19 and other respiratory illnesses, addressing global health needs.