A machine learning COVID-19 mass screening based on symptoms and a simple olfactory test

Area of Science:

• Infectious Diseases
• Medical Technology
• Data Science

Background:

• Early detection and rapid testing are crucial for controlling COVID-19 transmission.
• Olfactory dysfunction is a prevalent and often early symptom of COVID-19.

Purpose of the Study:

• To develop a machine learning predictive tool for COVID-19 detection.
• The tool integrates patient symptoms and a simple olfactory test using aromatized hydroalcoholic gel.

Main Methods:

• A multi-center, population-based prospective study in Reus, Spain.
• Included 519 patients undergoing PCR testing for COVID-19 symptoms or contact tracing.
• A classification tree model was developed using sex, age, symptoms, and olfactory test results.

Main Results:

• The machine learning model achieved a sensitivity of 0.97 (95% CI 0.91-0.99).
• Specificity was 0.39 (95% CI 0.34-0.44), with an AUC of 0.87 (95% CI 0.83-0.92).
• The model demonstrated high accuracy in predicting COVID-19.

Conclusions:

• The developed machine learning model shows promise as a mass screening tool for COVID-19.
• Integrating olfactory testing with symptom data enhances predictive capabilities.
• This approach supports efficient COVID-19 control strategies.