Infectious Diseases 2027

Background Antimicrobial Resistance (AMR), one of the top 10 global health threats, poses a critical public health challenge with a projected 10 million deaths annually by 2050 if left unchecked1,2. In 2019, 1.27 million deaths were attributed to AMR, with data gaps in low- and middle-income countries (LMICs) impeding global efforts 3,4. Fragmented databases limit the understanding of regional AMR burdens and hinder the development of data-driven interventions. For pharmaceutical companies, these gaps stifle research and drug development. Datasets like the Pfizer-ATLAS program are essential for real-time tracking of antibiotic resistance and for building predictive models to forecast AMR. Machine learning (ML)- based tools offer potential solutions for support clinical decision-making, personalized antibiotic prescriptions and AMR surveillance. Methods The microBIS platform combines a web server, AI/ML processing, and a user-friendly client interface. The MERN Stack workflow was used for web development and Python-based tools like Random Forest and PyBact for AI/ML modeling. Biochemical test results simulated and the Pfizer ATLAS datasets were used to train the platform's models to identify bacteria and predict AMR. Results The microBIS app offers intuitive interfaces for bacterial identification and AMR prediction. It uses biochemical test panels (8 or 20 tests) for bacteria identification, achieving a 79% accuracy with the 20-panel and 31% with the 8-panel. When predicting the top 5 bacterial species, accuracy increased to 99% for the 20-panel and 74% for the 8-panel. Certain tests, like citrate production, showed higher predictive value. For AMR prediction, the Machine Learning (ML) model trained on the Pfizer ATLAS dataset achieved 56% accuracy in predicting antimicrobial susceptibility profiles. Key predictors included bacterial phenotypes and species. Conclusions microBIS bridges the gap between accessible biochemical assays and advanced ML tools, offering a cost-effective, scalable solution for bacterial identification and AMR prediction. It empowers healthcare providers and researchers in resource-limited settings with rapid diagnostic capabilities, enhancing clinical decision-making and AMR surveillance. By integrating AMR datasets and predictive analytics, microBIS enables better care delivery and supports the global fight against AMR. References 1. Feldgarden, M., et al. (2021). AMRFinderPlus and the Reference Gene Catalog facilitate examination of the genomic links among antimicrobial resistance, stress response, and virulence. Scientific Reports, 11(1), 12728. https://doi.org/10.1038/s41598-021-91456-0

David Otieno is a Clinical Research Scientist at the Kenya Medical Research Institute (KEMRI), where he conducts research on infectious diseases with a focus on antimicrobial resistance, microbial genomics, and host–microbiome interactions. He has extensive experience in clinical and laboratory research, including molecular biology, microbiology, genomics, biobanking, and epidemiological studies. David has contributed to multidisciplinary research projects involving infectious disease surveillance, antimicrobial resistance, and translational health research, collaborating with national and international partners.He is currently pursuing research on investigating how probiotics reduce antibiotic resistance gene carriage in infants' gut microbiomes using functional omics approaches. His research aims to generate evidence that informs strategies to combat antimicrobial resistance and improve global health. David is passionate about translating scientific discoveries into
public health impact, strengthening research capacity in Africa, and fostering collaborations that advance infectious disease research and innovation.