Distinct Laboratory Test Result Profiles Between SARS-CoV-2 and Seasonable Influenza Infected Patients.

Area of Science:

• Medical Diagnostics
• Infectious Diseases
• Machine Learning in Healthcare

Background:

• Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and influenza viruses present with similar symptoms, necessitating distinct treatment and management.
• Differentiating between these pathogens in emergency departments (ED) is crucial for appropriate patient care.

Purpose of the Study:

• To investigate differences in laboratory test result profiles between SARS-CoV-2 and influenza infected patients presenting to the ED.
• To develop and evaluate machine learning models for differentiating these viral infections based on laboratory data.

Main Methods:

• Retrospective analysis of laboratory test results and demographic data from 723 influenza and 1,281 SARS-CoV-2 positive ED patients.
• Development and evaluation of four machine learning models, including a random forest (RF) model, to differentiate viral infections.
• Utilized Shapley Additive Explanations (SHAP) for visualizing laboratory test contributions to differentiation.

Main Results:

• A random forest model utilizing 15 blood tests, age, gender, and race effectively discriminated between SARS-CoV-2 and influenza infections.
• The RF model achieved an area under the receiver operating characteristic curve (AUC) of 0.90 in the test set.
• Serum total calcium level was the most significant laboratory test for differentiating the two infections.

Conclusions:

• A laboratory test-based random forest model was successfully developed to differentiate SARS-CoV-2 from influenza.
• This model shows potential utility in preparing for overlapping COVID-19 resurgence and seasonal influenza outbreaks.
• The findings highlight the role of specific laboratory markers in distinguishing common respiratory viral infections.