Yes, you already know that AMR is projected to cause 10 million deaths annually by 2050 without effective interventions [1]. It is also projected that, by the same ominous 2050, AMR could lead to cumulative losses to the tune of $100 trillion [2]. These catastrophic projections are the best points to start this essay. 

While it is understandable that the economic valuation of the AMR burden is in the US dollars, often we observe that data, statistics, outlook, and economic costs of the problem are cited in ways that reflect the interests of high-income countries. To be clear, AMR is a global problem. However, AMR weighs heavier on low and medium income countries. Africa, for example, has the highest mortality rate due to AMR with 27.3 deaths per 100,000 people [3].

LMICs are not “dollar economies”. They face different challenges. There is a stark infrastructural deficit, poor healthcare access, health inequities, and limited government financing. Because of these peculiar challenges, a one-size-fits-all solution (as we see with the dollarized economic framing) might undermine local realities.

Without solid traditional infrastructure, technology – especially Artificial Intelligence – presents an opportunity that can be easily leveraged by governments and investors in LMICs.

Enter Artificial Intelligence

Artificial Intelligence (AI) has taken the world by storm. AI has found applications in virtually every sector — healthcare included. From GPT-styled language models, offline AI-powered smartphone diagnostic tools, and predictive analysis, this essay will explore practical ways in which AI can be leveraged to improve health outcomes in low and medium income countries. 

Combatting Antimicrobial Resistance in Resource-Limited Countries

In resource-limited settings, poor ICP practices, inadequate laboratory capacity and diagnostic tools, and weak policies on antibiotic usage pose the greatest challenges. Failure of governments to address these problems have led to an overreliance on antibiotics. This overreliance is one of the chief drivers of antibiotic resistance. Today, many big pharmaceutical companies have withdrawn from antibiotic research and development because there is little incentive to pursue new drugs [4]. This has left LMICs in a precarious state. With the cost of developing new antibiotics pegged at about $1.2 billion, addressing the underlying challenges is the best bet. 

The remainder of this essay will focus on scalable approaches to leveraging AI in mitigating AMR burden in LMICs. 

Where AI Meets AMR

Having stated the problem and approach, here are ways that AI can improve health outcomes:

Improving Antibiotic Susceptibility Testing

Disk diffusion still reigns supreme for antimicrobial susceptibility testing. This cheap and simple method is, however, not without its failings. Detecting susceptibility and resistance can shape clinical decisions and lessen economic burdens on the patient especially in a country with a significant percentage of out-of-pocket [5] payments. Artificial Intelligence can improve access to reliable antimicrobial susceptibility tests through high-powered image processing. In particular, Medecins Sans Frontiers developed Antibiogo, an AI-powered smartphone app capable of providing highly accurate susceptibility results [6]. Interestingly, this tool, which has been deployed in some African countries, is fully operational offline. This allows it to evade the bottleneck of internet connectivity and power outages, while maintaining prompt analysis. 

Predictive Analysis and Clinical Decision Support Systems

In the book, Everything is Tuberculosis, John Green highlights the horrifying health disparities between high income nations and LMICs in tuberculosis treatment. Antibiotic resistance in Mycobacterium tuberculosis is very well documented [7]. But AI can be the game-changer. Several predictive analysis tools like GenTB and Mykrobe have been designed to analyze genome sequence data [8]. These tools have been trained on large datasets of TB isolates with resistant profiles. GenTB, in particular, uses Random Forest and Wide and Deep Neural Network to investigate genome sequences associated with resistance. In simpler terms, these tools can rapidly predict resistance probabilities to about 13 anti-TB drugs which can help inform clinical decisions, ensure targeted interventions, and improve health outcomes. AI can also be deployed to monitor online search entries for keywords that can signify a public health crisis. 

Streamlining Infection Prevention and Control (IPC) in Hospitals

Poorly equipped, poorly funded, and overburdened, hospitals in LMICs can serve as an infectious hotspot without standard IPC. A recent study has shown that implementing robust IPC programs in resource-limited countries can prevent 337,000 deaths per year [9]. Studies have also revealed that low-income and lower-middle income countries had the highest prevalence of hospital acquired infections at 37% and 22% respectively [10]. Additionally, 1 in 3 cases of hospital acquired infections are caused by resistant bacteria [11]. Infection Prevention and Control is lauded as one of the assured ways of breaking the AMR chain especially in resource-limited countries [12]. Artificial Intelligence can be deployed in hospitals to streamline IPC in a number of ways. For instance, in high-risk units like ICU or O&G wards, AI-powered sensors can be used to monitor environmental conditions like temperature, humidity, air quality, and even hand hygiene compliance. The use of computerized vision sensors to monitor handwashing routines has been test-run at Lucile Packard [13]. These data are continually fed to a central server in real time to ensure prompt response to infection and disease outbreak.

Strengthening Health Policy

Policymakers are one of the central players in the fight against antimicrobial resistance. While LMICs have set up frameworks against AMR (case in point, Nigeria’s National Action Plan 2.0) in recent times, there is still laxity in institutional adoption. Largely, this is because AMR is not generally perceived as a crisis in LMICs. An assessment of transparency and accountability of AMR NAPs in 15 African countries show that while many have drafted Action Plans, only one country had made funding allocations [14].

AI can inform policies by identifying health gaps like drugs stock-out rates, counterfeit medicines, and disease patterns. Local data obtained can be visualized by AI-powered dashboards making it easier for policy makers to design intervention strategies and draft budgets. This is not the only way AI can come into play. Recently, some researchers developed a language model similar to ChatGPT. This tool, aptly called AMR-Policy GPT [15], is an AI model designed to provide localized responses to AMR searches. Responses are grounded in the local realities and as such can help in forming realistic policies.

But There Are Some Limitations 

This essay has focused on AI solutions that are actionable, but it’s difficult to ignore the challenges that stand against their adoption. This section features some of the staunch challenges and recommends some workarounds.

Limited Digital Infrastructure

Digital Infrastructure is not fully developed in resource-limited communities. There are inadequate computer systems, poor network connectivity, and even electricity challenges. To solve this challenge, the government could explore solar solutions. Some hospitals already operate on solar-powered energy, and through collaboration with investors, this can be scaled into a nationwide action [16]. The connectivity issue has already been solved by tools like Antibiogo. For resource-limited communities, running AI-powered offline diagnostic tools on smartphones can solve this problem. The government has to invest in setting up solid digital infrastructure to support advanced procedures like genome sequencing.

 Algorithmic Bias

The earlier part of this essay briefly addressed a dollarization of the AMR problem. The term simply attempts to express that most statistics and solutions for AMR are collected or designed with developed countries in mind. Similarly, AI tools trained on data from first-world countries may not offer localized interventions in LMICs. This bias can be addressed by building local datasets that are reflective of the target demography. For this, health and research institutes must collaborate in ensuring that data is collected and stored digitally for easy analysis. This would be easier for urban clinics that already have electronic health records.

Human Capacity 

AI tools require tech-savvy hands. To ensure that the tools work to their fullest capacity, the government must work to ensure that health workers are trained to handle these digital tools. The training could simply be integrated into house job or internship trainings for young professionals or as short presentation workshops for older professionals.

Ethical Concerns

Ethics is a grey area. With new technology, there is always the issue of data privacy and safety. With deep-seated mistrust for the government in LMICs like Nigeria, convincing the populace to consent to using their data to train AI models might receive some pushback. Additionally, there could be some confusion regarding accountability. In the case that an AI tool proffers a wrong suggestion and treatment fails (when it’s deployed as a clinical decision support system), who should be held accountable? The government, healthcare professional, or even the developer of the tool itself. These grey areas can discourage full adoption of AI solutions. The government must clearly spell out guidelines for data collection, storage, and usage.

Conclusion 

The truth of the matter is that the AMR burden rests disproportionately on LMICs. Resolving the problem requires a multimodal and multi-sectoral approach. Of all the possible solutions — from grassroots reorientation, IPC, WASH, policy development, biosurveillance, data analysis — Artificial Intelligence can be applied.

By harnessing AI, LMICs can scale traditional infrastructural deficits and design solutions that are tailored to their own needs. AI can obviously be applied in even more cases to improve health outcomes (drug discovery is one), but this essay is an attempt to present the most feasible solutions.

Admittedly, some of these solutions — particularly that on predictive analysis and biosurveillance — would require serious work to set into motion. But they can be done if the stakeholders are convinced of the need to. Thankfully, AI can help present complex data and trends with compelling visuals.

References

1. O’Neill, J. (2016). Tackling drug-resistant infections globally: Final report and recommendations (Review on Antimicrobial Resistance). AMR Review.

2. O’Neill, J. (2016). Economic impact projections (Review on Antimicrobial Resistance). In Tackling drug-resistant infections globally: Final report and recommendations. AMR Review

3. Africa CDC. (n.d.). Antimicrobial resistance: New report warns of growing threat. https://africacdc.org/news-item/antimicrobial-resistance-new-report-warns-of-growing-threat/ 

4. Holmes, A. H., Moore, L. S. P., Sundsfjord, A., Steinbakk, M., Regmi, S., Karkey, A., Guerin, P. J., & Piddock, L. J. V. (2015). Understanding the mechanisms and drivers of antimicrobial resistance. Journal of Antimicrobial Chemotherapy, 70(6), 1604–1617. https://academic.oup.com/jac/article/70/6/1604/728792 

5. Kankaya, E. A., & Yilmaz, E. (2024). Financial protection and antimicrobial resistance: A global assessment. Frontiers in Public Health, 12, Article 11871602. https://pmc.ncbi.nlm.nih.gov/articles/PMC11871602/ 

6. Médecins Sans Frontières (MSF) Foundation. (2021–present). Antibiogo: Diagnostic and decision-support tool. MSF Foundation

7. World Health Organization. (2023). Global tuberculosis report 2023: Drug-resistant TB burden. https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2023/tb-disease-burden/1-3-drug-resistant-tb 

8. World Health Organization Regional Office for Europe. (n.d.). Improving infection prevention and control to prevent the spread of antimicrobial resistance. https://www.who.int/europe/activities/improving-infection-prevention-and-control-to-prevent-the-spread-of-antimicrobial-resistance 

9. Lewnard, J. A., Charani, E., Gleason, A., Hsu, L. Y., Khan, W. A., Karkey, A., Chandler, C. I. R., Mashe, T., Khan, E. A., Bulabula, A. N. H., Donado-Godoy, P., & Laxminarayan, R. (2024). Burden of bacterial antimicrobial resistance in low-income and middle-income countries avertible by existing interventions: An evidence review and modelling analysis. The Lancet, 403(10442), 2439–2454. https://doi.org/10.1016/S0140-6736(24)00862-6 

10. Odoom, A., Tetteh-Quarcoo, P. B., & Donkor, E. S. (2025). Prevalence of hospital-acquired infections in low- and middle-income countries: Systematic review and meta-analysis. Asia Pacific Journal of Public Health, 37(5), 448–466. https://doi.org/10.1177/10105395251338002 

11. European Centre for Disease Prevention and Control. (n.d.). Each year, 4.3 million patients in hospitals in the EU/EEA are affected by healthcare-associated infections. https://www.ecdc.europa.eu/ 

12. Harant, A., Gandra, S., Storr, J., & Senoner, T. (2022). Assessing transparency and accountability of national action plans on antimicrobial resistance in 15 African countries. Antimicrobial Resistance & Infection Control, 11(1), 15.

13. Haque, A., Guo, M., & Stanford Artificial Intelligence Lab. (2016–2018). Towards vision-based smart hospitals [Blog post]. Stanford AI Lab / Lucile Packard Children’s Hospital.

14. Gröschel, M. I., Owens, M., Freschi, L., et al. (2021). GenTB: A user-friendly genome-based predictor for tuberculosis resistance powered by machine learning. Genome Medicine, 13, 138. https://gentb.hms.harvard.edu 

15. Chen, C., Li, S. L., So, A. D., Xu, Y. Y., Guo, Z. F., Wang, X., Graham, D. W., & Zhu, Y.-G. (2025). Using large language models to assist antimicrobial resistance policy development: Integrating the environment into health protection planning. Environmental Science & Technology, 59(2), 1243–1252. https://doi.org/10.1021/acs.est.4c0784216. Godwin, A. (2024). Solar relief for hospitals in a power crisis. The Guardian (Nigeria). https://guardian.ng/opinion/solar-relief-for-hospitals-in-power-crisis/

FIRST POSITION : WAAW Essay Competition

Name: Abdulbaseet Yusuff

University: Usman Danfodio University 

Program: Pharmacy