Artificial intelligence has moved convincingly beyond proof-of-concept in AMR diagnostics and discovery, but its path to the clinic now depends less on further algorithmic refinement than on prospective validation, equitable data representation, and interpretability standards that clinicians can reasonably trust.
Abstract
Background: Antimicrobial resistance (AMR) is projected to contribute to millions of deaths in the coming decades, and the conventional antibiotic-discovery pipeline has, by most accounts, not kept pace with it. Artificial intelligence (AI) is frequently proposed as a corrective, though whether that promise has translated into demonstrable clinical benefit is less often examined directly.
Methods: We conducted a narrative-systematic review of peer-reviewed and preprint literature on AI applications in AMR diagnostics and antimicrobial discovery, searching PubMed/MEDLINE, Scopus, Web of Science, and the Cochrane Library through mid-2026, and organized findings across four domains: rapid phenotypic diagnostics, genomic and metagenomic resistome prediction, explainable AI, and de novo drug design.
Results: AI-enabled diagnostics reduced susceptibility-testing turnaround from a conventional 36–72 hours to under 2–4 hours in several platforms; genomic language models such as DNABERT outperformed conventional classifiers by 12–18% in resistance-gene classification; explainable AI methods, SHAP in particular, linked model predictions to known resistance mechanisms; and generative frameworks yielded antimicrobial peptide candidates with confirmed in vitro and in vivo activity. Nearly all of this evidence, however, derives from retrospective, single-center validation, and no AI-based AMR tool has yet secured regulatory clearance anywhere.
Conclusion: AI has moved convincingly beyond proof-of-concept in AMR diagnostics and discovery, but its path to the clinic now depends less on further algorithmic refinement than on prospective validation, equitable data representation, and interpretability standards that clinicians can reasonably trust.
Background and Objectives: Antimicrobial resistance (AMR) is a major challenge, particularly in intensive care units, where broad-spectrum therapy is often initiated before microbiological confirmation. Artificial intelligence (AI) may improve AMR prediction, but its clinical value depends on integration with bioengineering-enabled digital microbiology. This narrative review examines how AI, bioengineering platforms and digital microbiology can support AMR prediction, clinical decision support and antimicrobial stewardship across the sample-to-decision pipeline. Materials and Methods: A targeted narrative review was conducted using PubMed/MEDLINE and Google Scholar. Publications from 2020 onward were prioritized, while earlier seminal studies, methodological frameworks and regulatory documents were included when relevant. Evidence was synthesized across AI-based resistance prediction, antimicrobial stewardship, digital microbiology and bioengineering technologies. Results: AI and machine-learning approaches showed promising performance in patient-level resistance prediction, pathogen-level susceptibility prediction and antimicrobial stewardship. For example, model discrimination reached an AUROC of 0.936 for carbapenem-resistant Klebsiella pneumoniae prediction, while model-guided empirical therapy in Enterobacterales bloodstream infections could have increased active beta-lactam therapy from 70% to 79%. However, most evidence remains retrospective and single-centre, with limited external or prospective validation. Conclusions: AI has considerable potential to support AMR prediction and antimicrobial stewardship, but current evidence primarily demonstrates technical feasibility rather than established clinical effectiveness. Broader implementation will require rigorous validation, integration into clinical workflows, continuous monitoring and demonstration of clinical benefit.
Oana Frandeș, Leonard Azamfirei, Oana-Elena Branea et al.· Medicina· 0 citations
Antimicrobial resistance (AMR) represents one of the most critical global public health threats of the contemporary era, contributing to millions of deaths and substantial morbidity worldwide. Artificial intelligence and machine learning (AI/ML) have been increasingly applied to AMR detection, prediction, and management, demonstrating promising results in controlled research settings. However, clinical translation remains substantially constrained by fundamental methodological challenges, particularly data heterogeneity and algorithmic bias, whose extent and consequences are not yet adequately characterised.
A systematic review with descriptive synthesis was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines. Searches were performed across PubMed/MEDLINE, Embase, Scopus, Web of Science, IEEE Xplore, and the ACM Digital Library, covering studies published between January 2020 and April 2026. Eligible studies applied AI/ML approaches to AMR prediction using whole-genome sequencing (WGS), antimicrobial susceptibility testing (AST), electronic health records (EHRs), surveillance datasets, and spectral data. Data were extracted using a standardized charting form and synthesized using descriptive and thematic methods. No meta-analysis was conducted owing to substantial methodological heterogeneity across included studies.
A total of 68 studies published between 2020 and 2026 were included, of which 21 underwent detailed analysis. Studies demonstrated substantial heterogeneity across data modalities, patient populations, laboratory practices, and geographic settings. WGS-based approaches were most frequently represented, followed by AST- and EHR-based models. While many models achieved strong internal performance, generalizability across external settings remained limited. Key sources of heterogeneity included variability in data modalities, laboratory protocols, population composition, and temporal and geographic distribution. Major forms of bias identified included sampling bias, label inconsistency, structural confounding, and clinical context bias. Mitigation strategies demonstrated partial and context-dependent improvements.
Data heterogeneity and algorithmic bias are the primary constraints on AI/ML-based AMR prediction, not algorithmic sophistication. The strong internal performance metrics reported across the literature do not reliably generalize to diverse populations, geographic settings, or clinical environments. Meaningful progress requires a deliberate shift in research priorities toward globally representative datasets, harmonized laboratory standards, mandatory external validation, and equitable model development. The proposed Heterogeneity Mitigation Framework offers a structured, evidence-grounded approach to these challenges; its empirical validation in diverse real-world settings is the most important next step for the field.
Although this review was not prospectively registered in PROSPERO, it was conducted using predefined eligibility criteria, structured data extraction, and transparent reporting in accordance with PRISMA 2020 guidelines.
J. Kiazolu· BMC Medical Informatics and...· 0 citations
Abstract Antimicrobial resistance (AMR) poses a significant global public health threat, and efforts to mitigate it have been aided by artificial intelligence (AI) methods. Key areas of applications include rapid diagnostics, drug discovery and repurposing, surveillance and predictive modelling, and antibiotic stewardship. This review aims to summarize key literature on AI applications for mitigating AMR including original research articles from PubMed and Scopus published from October 2024 to 2025 to summarize and find out the way ahead for successful application of AI. The key search terms included were AMR, AI and applications such as diagnostics, drug discovery and repurposing, surveillance, predictive modelling and stewardship. The data used for these applications, the techniques applied and the predictive targets have undergone significant expansion, along with the focus on model deployment, external validation and model interpretability. Although more complex models, such as neural networks and transformers, have been experimented with, classical machine learning (ML) models still dominate the AMR prediction space, while large language models are also being tested for prediction, as well as antibiotic stewardship. For drug discovery, data mining for antimicrobial peptides from different sources is a major application. Predictive modelling using next-generation sequencing data has been the most studied. The application of AI/ML to large and complex data from multiple sources could provide a promising arena for developing clinically translational tools. With more data availability, regulatory measures, real-world validation and transparency, there is scope for responsibly integrating innovative technology into clinical practice.
Swetha Valavarasu, S. Marathe, Sanjay Kochar et al.· JAC-Antimicrobial Resistance· 0 citations
Antimicrobial resistance (AMR) represents one of the most critical global threats to public health; it was associated with an estimated 4.95 million deaths in 2019 [1] and is projected to claim up to 10 million lives annually by 2050 if no effective interventions are implemented [2]. The rapid dissemination of multidrug-resistant (MDR) bacterial strains continues to render first-line and last-resort antibiotics ineffective, outpacing traditional drug discovery pipelines. The rapid dissemination of multidrug-resistant (MDR) bacterial strains continues to render first-line and last-resort antibiotics ineffective, outpacing traditional drug discovery pipelines [3]. Historically, the development of new antibacterial agents relied on modifying existing chemical classes, a strategy that is increasingly vulnerable to rapid selection of resistance under clinical pressure [4]. To overcome this bottleneck, the modern research paradigm is shifting from single-molecule investigations to systems-level analyses [5]. The convergence of high-throughput multi-omics profiling spanning genomics, transcriptomics, proteomics, and metabolomics with advanced artificial intelligence (AI) and machine learning (ML) architectures offers a powerful framework for deciphering the complex molecular underpinnings of resistance and accelerating target-directed drug discovery [6].
Monochura Saha, Mehra Smriti, Priyanka Kumari· Advances in Medical Sciences...· 0 citations
Antimicrobial resistance (AMR) is currently one of the leading global health threats. The evolution of drug-resistant bacterial pathogens is rapid, and there is a growing number of bacterial pathogens that have developed resistance to multiple antibiotics, and the rate at which new antibiotics are being developed is lagging far behind these two issues. In addition, historical drug discovery processes rely on conducting traditional in vitro-based studies to determine new antibiotics to use in practice. However, this process is becoming increasingly constricted due to high costs of conducting traditional in vitro research, lengthy timeframes to bring products to market, a high attrition rate of research projects in traditional wet laboratory environments, and the need for more effective and efficient ways of developing new drugs. For this reason, Drug discovery continues to evolve from a wet lab-based approach to an in silico (i.e., computational) based approach, which takes advantage of the advances made in various fields, such as bacterial genomics, structural bioinformatics, machine learning (ML), and systems biology to enable researchers to rationally design, discover, and develop new antibiotics to combat drug-resistant pathogens. This review aims to provide a comprehensive and critical overview of contemporary in silico antibiotic discovery strategies and their potential to accelerate the development of novel, resistance-resilient, and clinically relevant antimicrobial agents. It examines genome-informed approaches ranging from genomic data generation, resistome analysis, and computational target identification to structure-based drug design, ligand-based and fragment-based discovery, drug repurposing, and the expanding applications of ML and artificial intelligence (AI) in activity prediction, de novo antibiotic design, and resistance evolution modeling. The review also highlights the importance of in silico ADMET prediction in lead optimization and discusses representative case studies demonstrating successful translation of computational predictions into experimental validation. Overall, the integration of digital-first, data-driven, and genome-guided discovery pipelines with experimental validation offers a powerful framework to overcome current challenges in antibiotic development and represents a promising strategy for addressing the global threat of AMR. Lastly, the review was conducted using a structured literature search across major biomedical and computational databases with emphasis on experimentally validated case studies and translational relevance.
Rohan Gupta, Pallavi Singh, Karthikeyan Ravi et al.· Frontiers in Bioinformatics· 0 citations
Background: Antimicrobial resistance (AMR) has narrowed the antibiotic arsenal for critically ill patients to a shrinking set of last-resort agents — polymyxins, carbapenems, glycopeptides, oxazolidinones, and glycylcyclines — and even these are now eroding under sustained selective pressure (Wahnou et al., 2026). Whether artificial intelligence (AI) can meaningfully close the gap between the emergence of resistance and its clinical recognition remains, we think, an open and worthwhile question.Methods: We conducted a narrative literature review of peer-reviewed articles, indexed preprints, and surveillance reports published between 2007 and 2026, retrieved from PubMed, Scopus, and Google Scholar using structured keyword combinations and supplemented by hand-searching of reference lists.Results: Resistance to last-resort agents arises through convergent mechanisms — mcr-mediated colistin resistance, carbapenemase production, van-gene peptidoglycan remodeling, and efflux-driven tigecycline resistance — disseminated via horizontal gene transfer across clinical, agricultural, and environmental reservoirs (Solanki & Kumar Das, 2024). AI-based annotation tools such as DeepARG and PLM-ARG now identify divergent resistance genes with precision exceeding 97% (Arango-Argoty et al., 2018; Wu et al., 2023), while optical and microfluidic platforms compress susceptibility testing from 24–72 hours to under 30 minutes in several validated systems (Liao et al., 2025). Generative models have also identified candidate antimicrobials, including halicin and abaucin (Bagdad & Miteva, 2024).Conclusion: AI is accelerating both detection and discovery, but database bias, limited interpretability, and pharmacokinetic translation failure remain substantial barriers. Integrating environmental metagenomic surveillance with clinical decision support, under a coordinated One Health framework, appears to be the most promising path toward preserving last-resort antibiotic efficacy.
A new method, called CW-Net, translates the reasoning process of an autonomous vehicle’s AI system into understandable concepts that explain its behavior.