Artificial intelligence technology supports global efforts to combat antimicrobial resistance

Antimicrobial resistance occurs when the microorganisms that cause infections, such as bacteria and viruses, change over time and stop responding to antibiotics.

It makes it harder to treat serious diseases such as HIV, tuberculosis and malaria, and increases the risk of serious illness, disease spread and death.

Antimicrobial resistance has a particular impact on low- and middle-income countries, where water quality is often poor and the spread of antimicrobial resistance in the environment through waste can be high.

In 2015, the World Health Organization (WHO) formulated a global action plan to coordinate efforts to combat antimicrobial resistance.

As a result, 194 WHO Member States have committed to developing ‘One Health’ national action plans (NAPs) on antimicrobial resistance.

The One Health model recognizes the interconnectedness between people, animals, plants and their shared environment.

However, insufficient logistical capacity, financing and poor access to necessary information may hamper informed NAP policymaking, especially in low- and middle-income countries.

Now, an international team of researchers, led by Professor Yong-Guan Zhu of the Chinese Academy of Sciences and Professor David Graham of the University of Durham in the UK, has created an artificial intelligence tool to fill critical gaps in knowledge needed to develop informal policies and help prepare National Action Plans.

The study results were published in the journal

The research team’s large language modeling tool, called AMR-Policy GPT, incorporates information from policy documents related to antimicrobial resistance from 146 countries.

It works in a similar way to established AI chatbots such as ChatGPT, but with a focal element that encourages more timely, accurate and contextually relevant information about AMR compared to more general chatbots.

We believe our prototype provides a valuable starting point for national action plans, especially for parts of the world that lack local data or infrastructure to enable integrated action against antimicrobial resistance.

Any solution to global health must be viewed holistically, and our tool will help guide policy development on antimicrobial resistance by increasing knowledge sharing between countries around the world, particularly regarding the spread of antimicrobial resistance in the environment.

Essentially, it provides decision-makers with well-documented information from all disciplines at their fingertips.

And with the ability to constantly update, our framework ensures that the chatbot tool remains current and effective.”

Professor David Graham from the Department of Biological Sciences, University of Durham

The researchers emphasize that the main goal of the AMR-Policy GPT is to be an “intelligent” source of information to assist in the policy-making process – like having an intelligent friend in the room, and is not intended for writing comprehensive NAPs.

Scientists will continue to develop the prototype and explore how it can be improved and expanded based on user feedback.

In the future, they would like to integrate even more scientific knowledge with policy information to create an improved GPT policy on antimicrobial resistance.

Professor Zhu said: “Given the vast and growing amount of information on antimicrobial resistance and its possible impact on policy, we believe that artificial intelligence is an excellent tool for integrating knowledge as well as for the initial distillation of understanding.”

Scientists from Shanghai Jiao Tong University in China and the Johns Hopkins Bloomberg School of Public Health in the US also participated in the study.