Abstract Rationale Optimizing biologic therapy in severe asthma relies on accurate early prediction of treatment response to prevent unnecessary delays, adverse effects, and costs. Existing models lack clinical guideline alignment and sufficient precision. We developed a quantum-inspired, multi-domain biomarker model that predicts response with 93% accuracy, aligned with GINA 2024 standards, to support personalized treatment decisions. Methods We analyzed data from 801 severe asthma patients across five biologic agents (Omalizumab, Mepolizumab, Benralizumab, Dupilumab, Tezepelumab). Using Quadratic Unconstrained Binary Optimization (QUBO), we systematically selected an optimal panel of 18 biomarkers from 76 candidates, including demographics, clinical measures, and engineered features. The feature set was used to train an XGBoost classifier with stratified 5-fold cross-validation, producing a continuous response score aligned with GINA thresholds. The model enables early (Month 3) response prediction and risk stratification with explicit confidence calibration. Results The model demonstrated an overall accuracy of 93. 4% (95% CI: 92. 2-94. 6%), with sensitivity 96. 8% and specificity 85. 1%. The AUC-ROC was 0. 969, indicating excellent discrimination. Key features included baseline Asthma Control Test (ACT) scores, eosinophil counts, exacerbation frequency, and demographic data, with consistent importance across configurations. The scoring system identified late responders, allowing therapy extension in multiple cases, supported by real-world follow-up data showing accurate predictions. The model’s clinical utility projected ∼60, 000 savings per responder by facilitating early therapy switches, decreasing ineffective treatment duration, and reducing exacerbation-related costs. Conclusions This innovative, guideline-aligned prediction model utilizing quantum-inspired feature selection and machine learning achieves state-of-the-art accuracy in forecasting biological response in severe asthma. It supports early, personalized treatment decisions, enables risk stratification with confidence calibration, and offers significant economic and clinical benefits. Prospective validation is ongoing, with integration into clinical practice anticipated to advance precision asthma management. This abstract is funded by: No
Salameh et al. (Fri,) studied this question.