Live
- Tigress Zeenat spotted in Purulia
- Collectors asked for proper assessment of crop loss
- Teachers are pillars of social development: Sudheer Reddy
- Sandhya Theater Stampede Incident: Rahul Ramakrishna Retracts Comments on Issue
- Ramanujan’s contributions to Mathematics hailed
- Garena Free Fire MAX Redeem Codes for December 2024: Unlock Skins, Weapons, and More
- Dr. Ponguleti Sudhakar Reddy Welcomes Union Minister Piyush Goyal in Bengaluru
- Why Christmas is Celebrated on December 25: The Link to Roman Festivals
- Tributes paid to G Venkataswamy (Kaka)
- Amit Shah should quit; demand MP, MLA
New study explores AI in developing antibiotics with less side-effects
The use of artificial intelligence (AI) has grown significantly, enabling applications such as content development, email proofreading, and driverless cars. New research explores its potential in developing improved antibiotics, with accurate and efficient prediction models.
New Delhi: The use of artificial intelligence (AI) has grown significantly, enabling applications such as content development, email proofreading, and driverless cars. New research explores its potential in developing improved antibiotics, with accurate and efficient prediction models.
Researchers from the University of Manitoba, Canada, deployed explainable AI (XAI) -- a branch of AI that provides a rationale for model judgments, increasingly being used by scholars to examine predictive AI models.
Though XAI can be applied in a variety of contexts, the team used it to develop antibiotics.
Despite AI's near-omnipotent use, many of its models act as "black boxes," making the decision-making process opaque. This can breed distrust, especially in vital domains such as drug discovery.
To overcome this, the team used XAI to train AI drug discovery models, particularly those that identify potential novel antibiotic candidates. Predictive models are essential given the pressing need for efficient antibiotics in the face of increasing resistance.
"AI is the way of the future in chemistry and drug discovery. It takes someone to lay the groundwork, and I think I'm doing that,” said Hunter Sturm, a graduate student of the University.
To anticipate biological effects, the scientists fed drug chemical databases into an AI model. An XAI model was then used to examine the precise molecular properties behind these predictions.
Fascinatingly, XAI discovered elements that human chemists would have missed, such as the fact that non-core structures in penicillin compounds are more important than the core itself.
To test predicted antibiotic compounds, the researchers collaborate with microbiology labs to improve AI models through the application of XAI insights.
Rebecca Davis, a professor of chemistry at the University of Manitoba, Canada noted that "AI causes a lot of distrust."
"Nevertheless, there's a better chance this technology will be accepted if we can ask AI to explain itself," she added.
The findings will be presented at the ongoing American Chemical Society (ACS) Fall 2024 meeting, being held from August 18-22.
