Equitable Performance of an AI-Based Digital Biomarker for Early Prediction of Childhood Asthma Exacerbations across Socioeconomic Strata 

Presenter: CHUNG-IL WI.

Socioeconomic status (SES) may affect the distribution of childhood asthma, raising the question of whether tools for early risk prediction perform consistently across socioeconomic groups. This retrospective cohort study evaluated whether SES modified the predictive performance of a natural language processing (NLP)-derived digital biomarker based on electronic health record data, which uses validated asthma phenotypes at age 3—Predetermined Asthma Criteria (PAC) and Asthma Predictive Index (API)—to identify children at increased risk of future asthma exacerbations (AE). Among 7,390 children (49% female; 82% White) in the Mayo Clinic Birth Cohort, SES at age 3 was classified using the individual-level HOUSES index in quartiles. Lower SES was associated with a higher prevalence of PAC+API-defined asthma (14% in Q1 vs. 10% in Q4; p<0.001), but SES was not independently associated with AE among children with asthma. Across HOUSES quartiles, the NLP-defined PAC+API phenotype consistently identified children with higher AE rates than PAC-only or API-only phenotypes (adjusted hazard ratios 2.22–2.97; p<0.05), except in Q2, with no evidence of SES-related effect modification (interaction p=0.68). These findings indicate consistent predictive performance of the NLP-derived biomarker across SES strata, supporting its potential for equitable early pediatric asthma risk stratification.

Very-Low-Dose Budesonide-Formoterol Maintenance and Reliever Therapy (MART) versus Fixed-Low-Dose Budesonide-Formoterol Plus as-Needed Salbutamol in Children 6-11 Years with Asthma: A Randomized Controlled Trial. 

Presenter: Sathya Srivatsav. 

Evidence supporting very-low-dose inhaled corticosteroid/long-acting β₂-agonist (ICS/LABA) Maintenance and Reliever Therapy (MART) in children aged 6–11 years requiring Step 3 treatment remains limited. This open-label, non-inferiority randomized controlled trial evaluated very-low-dose MART versus fixed-low-dose ICS/LABA plus as-needed SABA in newly diagnosed children aged 6–11 years requiring Step 3 therapy according to GINA guidelines. Of 98 randomized children, 94 (47 in each group) completed 3 months of follow-up; mean (SD) age was 8 (1.6) years and 71 (72.5%) were male. Change in ACQ-5 scores was similar with MART and fixed-dose ICS/LABA [1.61 (0.99) vs 1.59 (0.89); p=0.89]. The between-group mean difference was 0.02 [95% CI (-0.37 to 0.42)], remaining within the prespecified non-inferiority limit of 0.5. No significant differences were observed in FEV1 change, exacerbations, OCS use, emergency visits, or hospitalizations. Very-low-dose MART was therefore non-inferior to fixed-low-dose ICS/LABA plus as-needed SABA over 3 months. 

ERS Congress 2026, 5-9 September, Barcelona, Spain