Predicting laboratory aspirin resistance in Chinese stroke patients using machine learning models by GP1BA polymorphism

This study aims to use machine learning model to predict laboratory aspirin resistance (AR) in Chinese stroke patients by incorporating patient characteristics and single nucleotide polymorphisms of GP1BA and LTC4S. 2405 patients were analyzed to measure the Mutation frequency of GP1BA rs6065 and LTC4S rs730012. 112 patients with first-stroke arteriostenosis were prospectively enrolled to establish machine learning model. GP1BA rs6065 mutation frequency is 5.26% and LTC4S rs730012 is 14.78%. GP1BA rs6065 CT patients have more sensitivity to aspirin than CC genotype. Simple linear regression identified significant associations with age, smoking, HDL and GP1BA rs6065. Random forest (RF) and extreme gradient boosting (XGBoost) demonstrated predictive capabilities for AR. Findings suggest pre-identifying GP1BA rs6065 could optimize aspirin treatment, enabling personalized care and future research avenues.