Determination of prognostic markers for COVID-19 disease severity using routine blood tests and machine learning

Area of Science:

• * Medical Informatics
• * Clinical Pathology
• * Infectious Diseases

Background:

• * Identifying risk factors for COVID-19 severity is critical for patient care and resource allocation.
• * Current disease severity classification relies on non-standardized clinical parameters and blood tests, leading to conflicting data.
• * Machine learning (ML) offers a potential solution for developing standardized and accurate predictive models.

Purpose of the Study:

• * To construct and validate a machine learning (ML) model for predicting COVID-19 disease severity.
• * To identify key laboratory biomarkers associated with severe COVID-19 outcomes.

Main Methods:

• * A machine learning model was developed using electronic medical records and daily blood test results from 72 COVID-19 patients in Brazil.
• * Patients were diagnosed using RT-PCR and/or ELISA, with varying disease severity.
• * The model's predictive accuracy was assessed using the Receiver Operating Characteristic Area Under the Curve (ROC-AUC).

Main Results:

• * A combination of five laboratory biomarkers accurately predicted severe COVID-19 disease with a ROC-AUC of 0.80 ± 0.13.
• * The identified biomarkers include prothrombin activity, ferritin, serum iron, activated partial thromboplastin time (APTT), and monocytes.
• * The ML model demonstrated significant potential in predicting disease severity.

Conclusions:

• * The developed ML model effectively predicts COVID-19 severity using readily available laboratory data.
• * The identified biomarkers provide valuable insights into the pathophysiology of severe COVID-19.
• * This tool can aid in rationalizing clinical decision-making and optimizing patient care strategies.