Getting sick once is annoying. But getting sick again with the same exact condition is even worse. And it’s even worse when your prescribed medicine doesn’t do what it’s supposed to do. But the problem is bigger than that. Worldwide, about 700,000 people die each year as a result of resistance towards antibiotics. And if the problem of antibiotic resistance isn’t taken care of before 2050, these numbers will just continue to get larger, and larger.
The problem of Antibiotic Resistance
Antibiotic resistance, also known as antimicrobial resistance (AMR) is the phenomenon when bacteria, viruses, parasites or fungi evolve and don’t respond to antibiotics used to get rid of them. This is an issue because it makes many illnesses harder to treat, which leads to the risk of them spreading faster, and increasing in severity. AMR is a huge concern worldwide, WHO, the World Health Organization even says it’s “one of the top 10 global public health threats facing humanity.” So what is AI doing to better the situation?
Algorithms and Antibiotics
AI has been used to develop new antibiotics in a more efficient manner. For example, researchers at the Massachusetts Institute of Technology used an AI algorithm to more effectively find an antibiotic to kill the bacteria Acinetobacter Baumannii, a common hospital acquired pathogen that causes pneumonia and infections in other parts of the body. This bacteria is a concern to healthcare workers and patients because it is antibiotic resistant, and is the cause for over 300,000 deaths worldwide.
In order to make the antibiotic, researchers fed the molecular structure of the Acinetobacter Baumannii bacteria exposed to around 7,500 chemical compounds designed to stop bacterial growth, so that the algorithm could end up picking apart characteristics of each molecule that were associated with inhibition of bacterial growth. Once the algorithm processed everything, they introduced 6,680 new compounds to it, and after ending up with 240 compounds to test, they ended up with nine antibiotics, one of which proved to be incredibly effective against Acinetobacter Baumannii. This is a huge accomplishment because this could be a game changer when it comes to developing antibiotics. Developing antibiotics has always been a struggle for pharmaceutical companies and others involved in healthcare because of how expensive and time consuming it is. Even after numerous trials and effort, the antibiotics could end up being ineffective. AI speeds up this process and gives effective results in a shorter time frame.
Potential Drawbacks
Although AI has been proven to be quite useful in antibiotic development, there are still some bad things regarding antibiotic resistance. First is an occurrence called overfitting. This is when an AI model sees patterns between the data and outcomes that aren’t important to the development. This increases the amount of variables that affect the results, which causes the algorithm to make more inaccurate predictions. Another concern is described as the “black box problem.” AI models rarely give explanations for the predictions they make, and in the case that they are wrong, it becomes very difficult for scientists to learn anything from those predictions. More than that however, if in the case a prediction is wrong, it becomes an ethical issue because it’s very difficult for anyone involved to stand before the legal system or defend themselves.
To solve these issues, there are a number of methods that could possibly be implemented. The structure of these AI models could be changed by looking at other comparable transparent models. This could develop newer, more effective AI models that could explain the process behind their predictions. Regarding ethical issues, there are two approaches that can be taken: antibiotics with AI involvement could be strictly controlled, even if there’s no evidence that suggests risk. This is the approach a lot of Europe is using. The second approach to this would be to allow these antibiotics as long as there is no evidence that suggests a potential risk. As of now (In the US), it is still unclear as to which approach the FDA is going to take, as globally, there is no consensus on how they can and should be used.
The Future of Antibiotics
As of now, there is no one solution for antibiotic resistance, and it is still a debate as to whether or not AI will become mainstream in healthcare. These AI models will likely need lots more testing before they are released for public use, but nevertheless AI has the potential to change the future of healthcare, and it’s only a matter of time before everyone sees its effects.
Miethke, Et al. "Towards the Sustainable Discovery and Development of New Antibiotics." Nation=Al Library of Medicine, 19 Aug. 2021, www.ncbi.nlm.nih.gov/pmc/articles/PMC8374425/.
WHO . "Antimicrobial Resistance." World Health Organization, 17 Nov. 2021, www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance.
Trafton , Anne . "Using AI, Scientists Find a Drug That Could Combat Drug-resistant Infections." Massachusetts Institute of Technology, 25 May 2021, news.mit.edu/2023/using-ai-scientists-combat-drug-resistant-infections-0525#:~:text=courtesy%20of%20CDC-,Using%20an%20artificial%20intelligence%20algorithm%2C%20researchers%20at%20MIT%20and%20McMaster,for%20many%20drug%2Dresistant%20infections.
Dall, Chris. "Artificial Intelligence Discovers New Antibiotic Candidate." University of Minnesota, 26 May 2023, www.cidrap.umn.edu/antimicrobial-stewardship/artificial-intelligence-discovers-new-antibiotic-candidate.
Melo, M.C.R., Maasch, J.R.M.A. & de la Fuente-Nunez, C. Accelerating antibiotic discovery through artificial intelligence. Commun Biol 4, 1050 (2021). https://doi.org/10.1038/s42003-021-02586-0
Khan, Bangul et al. “Drawbacks of Artificial Intelligence and Their Potential Solutions in the Healthcare Sector.” Biomedical materials & devices (New York, N.Y.), 1-8. 8 Feb. 2023, doi:10.1007/s44174-023-00063-2
Writer : Manasvi Kattekola
Editor : Manav Desai
Comments