Irsan Hardi - Machine Learning Approach to Predict AXL Kinase

Journal article

T. R. Noviandy, Ghalieb Mutig Idroes, Irsan Hardi
Journal of Soft Computing and Data Mining, 2024

Cite

APA Click to copy
Noviandy, T. R., Idroes, G. M., & Hardi, I. (2024). Machine Learning Approach to Predict AXL Kinase Inhibitor Activity for Cancer Drug Discovery Using Bayesian Optimization-XGBoost. Journal of Soft Computing and Data Mining.

Chicago/Turabian Click to copy
Noviandy, T. R., Ghalieb Mutig Idroes, and Irsan Hardi. “Machine Learning Approach to Predict AXL Kinase Inhibitor Activity for Cancer Drug Discovery Using Bayesian Optimization-XGBoost.” Journal of Soft Computing and Data Mining (2024).

MLA Click to copy
Noviandy, T. R., et al. “Machine Learning Approach to Predict AXL Kinase Inhibitor Activity for Cancer Drug Discovery Using Bayesian Optimization-XGBoost.” Journal of Soft Computing and Data Mining, 2024.

BibTeX Click to copy

@article{t2024a,
  title = {Machine Learning Approach to Predict AXL Kinase Inhibitor Activity for Cancer Drug Discovery Using Bayesian Optimization-XGBoost},
  year = {2024},
  journal = {Journal of Soft Computing and Data Mining},
  author = {Noviandy, T. R. and Idroes, Ghalieb Mutig and Hardi, Irsan}
}

Abstract

This study aims to predictAXL kinase inhibitors utilizing aBayesian Optimization-XGBoost machine learning model. A dataset comprising 1074 compounds with IC50 values was collected from the ChEMBL database and molecular descriptors for each compound were calculated. The Bayesian Optimization-XGBoost model demonstrated superior performance in predicting AXL kinase inhibitors, achieving an accuracy of 86.24%, precision of 89.52%, recall of 89.52%, and an F1-score of 89.52%, outperforming other models such as LightGBM, Logistic Regression, k-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Naïve Bayes. This study underlines the importance of advanced machine learning techniques, particularly Bayesian Optimization-XGBoost, in predicting AXL kinase inhibitors, offering a promising approach for accelerating the early stages of drug discovery. Despite its success, the model's performance depends on the diversity and quality of the training data, and future work should focus on expanding the dataset and validating results with experimental studies. This computational method has the potential to streamline the drug development pipeline and contribute to the discovery of more effective cancer treatments.