Machine Learning in Genomic Analysis: ۲۰۲۵ Advances, Challenges, and the Path to Equity

Machine learning (ML) has transformed genomic analysis by extracting meaningful patterns from complex biological data, yet persistent challenges—including limited model interpretability, data imbalance, and population bias—threaten equitable translation into clinical and agricultural practice. This systematic review synthesizes evidence from ۲۰ state-of-the-art studies published in ۲۰۲۵, following PRISMA guidelines to examine ML applications in antimicrobial resistance (AMR), cancer diagnostics, and sustainable agriculture. Findings reveal remarkable advances: ML models achieve ۹۰-۹۹% accuracy in AMR prediction and cancer classification, while privacy-preserving frameworks like federated learning address data security concerns. Hybrid bio-inspired algorithms have reached ۱۰۰% accuracy in cancer diagnostics, setting new benchmarks for precision medicine. However, these successes are tempered by persistent challenges. Data imbalance skews performance, limited interpretability hinders clinical trust, and population bias remains entrenched, with approximately ۸۰% of genomic datasets derived from European ancestry—a disparity limiting global generalizability and risking health inequities. Critical trade-offs persist between accuracy and transparency, scalability and precision, and computational efficiency and sustainability. Future research must prioritize diverse population inclusion, interpretable AI frameworks, and sustainable computational strategies to ensure equitable progress in precision medicine and global health.

کلیدواژه ها:

machine learning ، genomics ، antimicrobial resistance ، cancer diagnostics ، privacy-preserving analysis ، equity ، sustainability