COVID-19 mortality prediction in Hungarian ICU settings implementing random forest algorithm

Area of Science:

• Infectious Diseases
• Critical Care Medicine
• Genomics

Background:

• Emerging SARS-CoV-2 variants of concern (VOCs) have altered Intensive Care Unit (ICU) demographics.
• The correlation between SARS-CoV-2 genotypes and ICU lethality requires further investigation.

Purpose of the Study:

• To analyze clinical and laboratory parameters in correlation with SARS-CoV-2 variant genotypes and ICU lethality.
• To develop a predictive model for SARS-CoV-2 lethality in ICU patients.

Main Methods:

• Retrospective study of 503 COVID-19 ICU patients (January 2021-November 2022) in Hungary.
• Implementation of a random forest (RF) model for lethality prediction in 649 ICU patients.
• Survival analysis comparing effects of hypertension (HT), diabetes mellitus (DM), and vaccination.

Main Results:

• Diabetes Mellitus (DM) was a significant mortality risk factor (OR: 1.55, p=0.025).
• Vaccination demonstrated a protective effect against mortality (p=0.028).
• The RF model achieved 81.42% accuracy in predicting lethality, with key predictors including PaO2/FiO2 ratio, lymphocyte count, and CTSS.

Conclusions:

• While Omicron variants may require less ICU treatment, survival rates for admitted patients have not proportionally increased.
• The developed RF model can aid in clinical decision-making for ICU COVID-19 patients.
• Understanding variant impact on ICU outcomes is crucial for patient management.