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

Area of Science:

• Genetics
• Cardiology
• Machine Learning

Background:

• Aspirin resistance (AR) is a clinical challenge in stroke patients.
• Genetic factors, including single nucleotide polymorphisms (SNPs) in genes like GP1BA and LTC4S, may influence AR.
• Personalized medicine approaches are needed to optimize antiplatelet therapy.

Purpose of the Study:

• To develop and validate a machine learning model for predicting laboratory aspirin resistance in Chinese stroke patients.
• To investigate the association of patient characteristics and SNPs in GP1BA (rs6065) and LTC4S (rs730012) with aspirin resistance.
• To explore the potential of GP1BA rs6065 genotype as a predictive marker for aspirin response.

Main Methods:

• Analysis of 2405 patients to determine mutation frequencies of GP1BA rs6065 and LTC4S rs730012.
• Prospective enrollment of 112 patients with first-stroke arteriostenosis for machine learning model development.
• Utilized simple linear regression, Random Forest (RF), and Extreme Gradient Boosting (XGBoost) for analysis and prediction.

Main Results:

• The mutation frequencies for GP1BA rs6065 and LTC4S rs730012 were 5.26% and 14.78%, respectively.
• GP1BA rs6065 CT genotype was associated with increased aspirin sensitivity compared to the CC genotype.
• Age, smoking, HDL levels, and GP1BA rs6065 were significant predictors identified through linear regression.
• RF and XGBoost models showed predictive capability for aspirin resistance.

Conclusions:

• Machine learning models can effectively predict aspirin resistance in stroke patients.
• The GP1BA rs6065 genotype is a potential biomarker for predicting aspirin response.
• Pre-identifying GP1BA rs6065 status can guide personalized aspirin treatment strategies for stroke patients.