Data Challenge: MDR-Predict: Fusing Genomic-Rich Artificial Intelligence Methods for Predicting Future Bacterial Multidrug Resistance Threats

Antibiotic resistance poses a critical global health threat, particularly due to the emergence of multidrug-resistant (MDR) bacterial pathogens that compromise the effectiveness of standard treatments. Pseudomonas aeruginosa, a high-priority ESKAPE pathogen, exhibits complex resistance mechanisms, necessitating innovative diagnostic strategies. This study proposes a comprehensive genomic feature analysis to identify and evaluate key sequence-based variables associated with antibiotic resistance in P. aeruginosa using the Pfizer-ATLAS whole-genome sequencing dataset.
By leveraging advanced feature engineering—including mutation profiles, sequence conservation, regulatory motifs, physicochemical properties, and expression-linked traits—we aim to define high-confidence MDR signatures. These features will be integrated into supervised learning models, including Support Vector Machines, Random Forests, logistic regression, and deep learning architectures such as deep neural networks, recurrent neural networks, and transformer-based models.
The objective is to develop a predictive framework capable of accurately classifying MDR variants with improved speed and precision compared to traditional susceptibility testing. Model performance will be rigorously validated using cross-validation and metrics like accuracy, F1-score, and AUC. Ultimately, this work will facilitate rapid, genomics-based AMR diagnostics, reduce dependence on culture-based methods, and pave the way for personalized antimicrobial therapy and real-time surveillance.