Unraveling Antibiotic Resistance in Escherichia coli: A Genomic Prediction Breakthrough

Escherichia coli has reached the point of antibiotic resistance, becoming a serious public health problem that renders treatments ineffective and can result in infections that are not treatable. Targeted therapy and reduced misuse of antibiotics could be hastened by early aggression of resistance patterns. In this work, we investigate the machine learning models that predict antibiotic resistance of the E. coli bacteria at the genomic and phenotypic levels. We used several machine learning algorithms. We evaluated the performance with accuracy, precision, recall and f1 score. Although numerous studies operate in this domain, our results indicate that XGBoost achieved the highest accuracy of 92.1%. The main novelty of our research is the feature selection strategy, optimization techniques of the model as well as the combination of multiple data to improve predictive performance. Unlike traditional statistical approaches, our method leverages advanced machine learning techniques to identify key resistance patterns effectively. The findings suggest that machine learning can serve as a reliable tool for predicting antibiotic resistance in clinical settings, helping to improve treatment decisions. Future work can focus on expanding the dataset and incorporating explainable AI techniques to enhance model interpretability.