Combining data from children’s saliva and blood samples and medical history can significantly improve predictions of future asthma attacks. This is according to the results of the new study, “Complementary Predictors for Asthma Attack Prediction in Children: Salivary Microbiome, Serum Inflammatory Mediators and Past Attack History.”
The study, which was published in the European Journal of Allergy and Clinical Immunology, was conducted through the SysPharmPediA and U-BIOPRED initiatives. It offers new hope for early intervention and personalized treatment strategies for pediatric asthma patients.
Researchers analyzed data from 154 school-aged children with asthma, integrating three key sources: the salivary microbiome, serum inflammatory mediators and past asthma attack history. The team used machine learning models — specifically, random forest algorithms — to develop predictive tools that could distinguish between children at risk and those not at risk of severe asthma attacks over a one-year period.
Although models based solely on past attack history or biological markers like salivary bacteria and inflammatory proteins achieved moderate accuracy (AUROCC ~0.7), the most effective model combined all three data types. This integrated approach reached an impressive AUROCC of 0.87 in the discovery phase and 0.84 in the replication phase, demonstrating strong predictive power.
Key biological indicators included three types of salivary bacteria — capnocytophaga, corynebacterium and cardiobacterium — alongside serum markers such as TIMP-4, VEGF and MIP-3α.
Researchers said the study underscores the importance of looking beyond traditional clinical history and incorporating biological data to better identify children at risk and tailor treatments accordingly.
Additionally, researchers noted the findings pave the way for future research into the role of the oral microbiome and immune system interactions in asthma, potentially transforming how pediatric asthma is managed.
