A Machine Learning Approach as an Aid for Early COVID-19 Detection

Area of Science:

• Infectious Diseases
• Machine Learning
• Public Health

Background:

• The COVID-19 pandemic caused by SARS-CoV-2 necessitated widespread public health measures, including lockdowns.
• Developing nations face challenges in accessing essential diagnostic resources for COVID-19.
• There is a need for accessible and scalable methods for COVID-19 detection.

Purpose of the Study:

• To develop and evaluate a machine learning approach for detecting COVID-19 infections based solely on self-reported symptoms.
• To assess the feasibility of an inexpensive and easily deployable diagnostic tool for COVID-19.
• To provide a potential solution for COVID-19 screening in resource-limited areas.

Main Methods:

• A machine learning model was trained using self-reported symptom data.
• The model's performance was evaluated using sensitivity, specificity, and ROC AUC metrics.
• The approach focused on symptom-based detection without requiring laboratory diagnostics.

Main Results:

• The best-performing model achieved a sensitivity of 0.752 and a specificity of 0.609.
• The receiver operating characteristic (ROC) curve area under the curve (AUC) was 0.728.
• These results indicate a promising performance for a symptom-based detection method.

Conclusions:

• Symptom-based machine learning models show potential for COVID-19 detection.
• This approach offers a cost-effective and scalable alternative or supplement to traditional diagnostics.
• Further research is warranted to refine these models for widespread application in COVID-19 screening.