Evolusi Patogen dan Resistensi Antimikroba: Integrasi Biologi Molekuler dan Kecerdasan Buatan

Abstract

Pathogen evolution and antimicrobial resistance (AMR) represent urgent global health threats, projected to cause 10 million deaths annually by 2050. Traditionally addressed through biological and clinical microbiology approaches, the limitations of conventional diagnostics have driven the integration of Artificial Intelligence (AI) and machine learning (ML) as revolutionary computational solutions. This review examines two co-equal themes in parallel: (1) biological mechanisms of pathogen evolution—including spontaneous mutation, horizontal gene transfer (HGT), genetic recombination, and antimicrobial selective pressure; and (2) AI/ML applications in prediction, surveillance, and control of pathogen evolution—encompassing Random Forest, Gradient Boosting, Support Vector Machine, Recurrent Neural Networks (LSTM), Convolutional Neural Networks (CNN), and multi-task deep learning architectures. Through a systematic review of literature from Scopus, ScienceDirect, IEEE Xplore, Emerald, and SINTA databases (2019–2025), this study demonstrates that ML models can predict AMR phenotypes from whole-genome sequencing (WGS) data with AUROCs of 0.77–0.98, while deep learning models have successfully identified high-risk SARS-CoV-2 mutations up to four months before the emergence of new Variants of Concern (VOCs). This integration of biological and computational approaches opens a new paradigm in pandemic preparedness and AMR control, with particular relevance for Indonesia as it builds a data-driven national AMR surveillance capacity.