Medicosnext
Dr. Shishir Pokhrel
He is an Assistant Professor of Microbiology at Pokhara Academy of Health Sciences, Nepal. A medical microbiologist with an MD in Microbiology, he is actively involved in undergraduate teaching, diagnostic microbiology, curriculum development, and antimicrobial stewardship. His research interests focus on antimicrobial resistance, and he has published peer-reviewed articles in medical journals.
Antimicrobial resistance (AMR) has been identified as one of the most serious global health threats in the 21st century. The emergence of resistant pathogens has been accelerated by the misuse of antimicrobials in human, veterinary, and agricultural sectors. This rampant misuse of antimicrobials has resulted in an increased burden of drug- resistant infections across the world. Therefore, the issue of AMR needs to be addressed not only by developing new antimicrobials but also by using antimicrobials in a rational and evidence-based manner.
Antimicrobial resistance (AMR) is one of the most serious global health threats of our time, and it is largely driven by the misuse of antimicrobials. Infections caused by resistant organisms result in prolonged illness, increased healthcare costs, and increased mortality. In this difficult scenario, artificial intelligence (AI) is being recognized as a powerful tool that can help improve antimicrobial use and antimicrobial stewardship.
Antimicrobial stewardship programs are designed to optimize antimicrobial use by ensuring that the patient receives the most appropriate antimicrobial agent, dose, route, and duration of treatment. However, antimicrobial stewardship programs are often limited by the lack of human resources, increased patient complexity, and the exponential increase in clinical and laboratory data. In this scenario, digital innovations, including artificial intelligence, have opened up new avenues to assist clinicians and microbiologists in making decisions.
Clinical decision support systems using AI can process enormous amounts of data such as patient demographics, clinical observations, microbiology lab reports, and local resistance data to provide the best possible antimicrobial therapy. By assisting in the selection of the right drug, in the right dose, and for the right duration, AI can help avoid the unnecessary use of broad-spectrum antibiotics and improve patient outcomes.
In addition to making prescribing decisions, AI can also help in predicting infection risk, identifying patients at high risk of resistant infections, and assisting in early de-escalation therapy. Predictive analytics models can use patient history data, comorbidities, previous antimicrobial exposure data, and hospital epidemiology data to predict the probability of resistance even before culture results are available. Early prediction of resistance can help in earlier initiation of effective therapy, which is a known predictor of reduced mortality in severe infections such as sepsis.
Recent studies from Algeria have shown that AI is already revolutionizing clinical microbiology. Machine learning (ML), deep learning (DL), and convolutional neural networks (CNNs) have been successfully applied in all areas of bacteriology, virology, parasitology, and mycology, including rapid pathogen detection and antimicrobial resistance prediction. These include AI-assisted analysis of COVID-19 RT-PCR tests, automated bacterial colony counting, malaria diagnosis from blood smears, and genomic
analysis for AMR prediction. AI has been successfully integrated with MALDI-TOF mass spectrometry, digital microscopy, and biosensors to provide rapid laboratory reporting and earlier optimization of antimicrobial therapy.
The use of AI in clinical microbiology laboratories is important because one of the factors that contribute to the misuse of antimicrobial agents is the delay in diagnosis. Conventional culture-based techniques, although accurate, are time-consuming and labor- intensive. AI-assisted diagnostic platforms can speed up the process of identifying pathogens and their resistance patterns, hence minimizing the use of empirical antibiotics. Moreover, image analysis and pattern recognition software decrease the subjectivity of human observers, hence increasing efficiency.
Despite the established potential, the actual application of artificial intelligence in clinical microbiology has been limited. Findings from a national survey of clinical microbiology laboratories in Italy suggest that only a quarter of the laboratories actually use AI or machine learning algorithms, mainly in bacteriology and virology. There was also a large gap in knowledge, and a very few laboratories actually using trained data scientists. The main obstacles to implementation were the lack of trained personnel, inadequate digital infrastructure, fragmented data systems, cost issues, and concerns about data privacy and trust in AI outputs.
These concerns are not limited to Italy but are indicative of broader challenges within the healthcare sector. The integration of AI in healthcare requires high-quality standardized data, adequate information technology infrastructure, and collaboration between clinicians, microbiologists, data scientists, and policymakers. Without these, AI tools may remain underutilized or generate biased and unreliable results.
However, there is hope. Almost all respondents showed interest in AI training, and many were eager to work together on AI projects. Large language models (LLMs) may not be in widespread use, but they were considered promising tools for data analysis, report writing, and workflow assistance. Most professionals also agreed that AI should be used to assist, not replace, human expertise.
This view highlights a guiding principle in the appropriate use of AI in healthcare: AI should be used as an assistive technology and not as an autonomous tool. Clinical acumen, context, and ethics remain fundamental to patient care. AI systems must, therefore, be transparent, explainable, and continuously monitored by humans to ensure patient safety and trust.
Because of the high burden of infectious diseases, the availability of over-the-counter antibiotics, and the low capacity for diagnosis, AMR is a serious public health issue for countries like Nepal. Although highly advanced AI technology is not prevalent, Nepal’s involvement in the WHO Global Antimicrobial Resistance and Use Surveillance System (GLASS) and the country’s antimicrobial stewardship program lay a foundation for future AI implementation.
In resource-constrained countries like Nepal, the progressive implementation of AI technology may provide great benefits if implemented according to the country’s needs and capabilities. Rather than carrying out highly advanced and expensive technologies, a stepwise strategy utilizing existing digital infrastructure may be more viable. Capacity building and staff training will be crucial to ensure long-term successful implementation. Even simple AI-assisted technologies like digital antibiograms, resistance trend analysis, and decision-support alerts within laboratory information systems could greatly improve
antimicrobial use. With gradual development of digital infrastructure, staff training, and governance, AI technology could become a viable and expandable solution in the Nepalese healthcare system.
Additionally, AI technology can aid in national and regional surveillance by combining data from various healthcare facilities to identify new trends of resistance and outbreaks. Real-time surveillance is essential for guiding public health responses, updating treatment guidelines, and allocating resources. When combined with the country’s AMR action plan, AI-assisted surveillance can improve readiness and response to the threat at both the local and national levels.
However, ethical and regulatory issues also need to be taken into consideration as the use of AI is on the rise. Matters pertaining to data privacy, bias, accountability, and equity are of utmost importance in a healthcare setting. It is essential that we have defined regulations and ethics to ensure that AI is used for the betterment of society as a whole, including the vulnerable population.
Conclusion
Artificial intelligence is not a substitute for clinical judgment but a reliable assistant. If used wisely, it has the potential to improve antimicrobial stewardship, promote rational use of antibiotics, and ensure that future generations are able to benefit from life-saving antimicrobials. It has the potential to become an integral part of our fight against antimicrobial resistance.
