A Predictive Model for the Development of Long COVID in Children

Area of Science:

• Pediatric Infectious Diseases
• Computational Medicine
• Epidemiology

Background:

• Long COVID poses significant risks for children, necessitating predictive tools for early intervention.
• Machine learning offers a promising approach to identify children at high risk of developing long-term complications after SARS-CoV-2 infection.

Purpose of the Study:

• To develop and validate a predictive algorithm for long COVID in pediatric patients post-SARS-CoV-2 infection.
• To identify demographic, clinical, and laboratory factors associated with long COVID development in children.

Main Methods:

• A cross-sectional study of 305 pediatric patients (1 month to 18 years) who recovered from acute SARS-CoV-2 infection.
• Development of two multivariate regression models incorporating various clinical and laboratory parameters.
• Analysis included demographic data, comorbidities, acute infection symptoms, vitamin D, zinc levels, complete blood count, and coagulation profile.

Main Results:

• Among 265 children analyzed, 52.0% developed long COVID.
• Model 1 (including non-hospitalized patients) identified obesity, allergic disorders, and vitamin D deficiency as significant predictors.
• Model 2 (hospitalized patients) identified fever, ≥3 acute symptoms, allergic conditions, thrombocytosis, neutrophilia, and altered prothrombin time as significant predictors.

Conclusions:

• Obesity, allergic pathology, fever, multiple acute symptoms, thrombocytosis, neutrophilia, altered prothrombin time, and low vitamin D levels are significant predictors of long COVID in children.
• The developed predictive model is highly acceptable and performs better than simple average predictions.
• These findings can guide clinical practice for timely detection and management of pediatric long COVID.