The world is very interested in generative artificial intelligence (GAI), and for good reason. People have a better idea of what AI can do thanks to platforms like ChatGPT that have shown what it can do. AI is also showing promise in the field of drug finding. AI is opening up new ways to do things, like creating new drug molecules and guessing how proteins are put together. It could speed up every step of the process, from finding targets to molecular models and guessing drug properties. AI has the potential to do more in drug creation than just speed up processes that have been slow in the past. It could lead to new ways to treat diseases that were once thought to be impossible to treat.
Changes for the better
In traditional ways of finding new drugs, big information sets are first screened. Hit-to-lead tuning and preliminary tests come next. It takes a lot of time and resources to do this whole process. The WHO says that the average cost of making a new drug is between US$43.4 million and US$4.2 billion, and that only 10% to 15% of clinical drug development projects are successful.
AI is a hopeful way forward because it helps people analyze huge datasets, find possible drug candidates, predict clinical results, and make the best decisions about how to plan clinical trials. A lot of important steps have already been taken with this technology. Exscientia made a big step forward in 2020 when the first drug chemical created by AI went into human clinical testing. Since then, DeepMind’s AlphaFold has helped us learn more about biology by guessing the shapes of more than 200 million proteins.
How well have AI methods worked so far? That question was looked into in a study from 2024, whose subject was “How successful are AI-discovered drugs in clinical trials?” It came to the conclusion that molecules made by AI have a success rate of 80% to 90%, which is a lot higher than the average success rate in the past. The study also found that AI-native biotechs and pharma partners have brought 75 molecules into the lab since 2015; as of 2023, 67 of these molecules were still being tested. AI-driven drug options from companies like Insilico Medicine are also making news as they move through clinical studies.
In this case, AI tools
AI technology gets a lot of attention, but the word itself isn’t always clear. Machine learning (ML), deep learning, natural language processing (NLP), and generative models are all types of AI methods. Each makes a unique contribution to a different stage of drug research.
Algorithms for machine learning (ML) use data to learn how to make choices and guesses. Researchers use ML a lot to guess how drugs will interact with their targets, look at biology data, and find the best ways to make drugs. Deep learning, a type of machine learning that uses neural networks to make predictions about protein shapes and interactions, is very important for understanding how drugs work. AlphaFold2 (from Google DeepMind) and ESM3 (from Evozyne, which was started by former Meta workers) use deep learning to guess the shapes of almost all known proteins. This changes how we think about how diseases work. Natural language processing, or NLP, lets computers understand and make sense of what people say. Scientists use NLP to get useful data from scientific papers, patents, and clinical study data. This helps them come up with new hypotheses and learn more about the world. Generative models are changing the way drugs are found by creating new chemical molecules with specific qualities that are wanted. This method can greatly speed up the process of finding good drug prospects.
Lastly, Large language models (LLMs) are changing the way drugs are found because they can use advanced natural language processing to look at and create complicated biology and chemical data. LLMs that have been trained on molecular data can make new structures that meet certain criteria, like binding affinity and selectivity, because they can use the medicinal chemistry principles that were stored in their training data.
Key places where AI is changing how drugs are made
During the finding phase, AI helps find targets and make sure they are correct by looking at omics data to locate genetic changes connected to diseases. Using generative models, it also speeds up the creation and improvement of drugs. Techbio companies like Exscientia and InSilico Medicine have made a big step forward by stating that the first medicine molecule created by AI will be tested on humans. AI also helps find biomarkers by looking at clinical and genetic data to find biomarkers for diagnosis, prediction, and treatment reaction. This makes personalized medicine possible. During initial research, AI systems can predict how harmful a drug option will be. This lowers the chance that the drug will fail in the later stages. They can also guess how drugs will act in the body and pick better candidates.
Several companies in the field of genetic medicine are using AI to speed up the creation of complicated biotherapeutics. AI-based in silico models are used to find and create the best payloads and vectors that are needed for genetic medicine. This mix of AI and genetic medicine has a lot of potential to make individual treatments better and open up new futures.
A number of businesses are also working on AI-powered gene therapy vectors that will make it easier to target specific cells with adeno-associated virus (AAV) vectors. This method not only makes gene treatments work better, but it also opens up new ways to treat many illnesses.
Companies like Asklepios Biopharmaceutical (AskBio), which was bought by Bayer, use AI in their gene therapy research to better understand how genes are controlled and find new regulatory sequences in genomes. Compared to normal biotech companies, Techbio’s method makes the process of making next-generation gene treatments more efficient and increases the chances of success.
Problems and a plan for how to solve them
Using AI in drug research comes with a number of difficulties and strategy steps. Setting up a good data plan is one of the most important things to think about. For building accurate AI models, it is important to have high-quality, varied, and reliable information. This is because the data these models are fed determines how well they work. It is very important to test AI predictions thoroughly using in vitro and in vivo methods to make sure they are accurate and useful in real life. Putting AI ideas into action and making them bigger is another big problem. Adding AI to current processes and making it available to many areas will take a lot of work and planning. AI models need to be constantly checked to make sure they stay correct and useful with new data. This means that model maintenance is always needed. People must be taken into account. It is important for R&D teams to work together because different science communities have different levels of interest and knowledge. There is still a need for awareness because many people are still not sure how AI will affect their field. Pharmaceutical businesses are fighting hard to hire highly sought-after professionals like computational and structural biologists because they need new skills.
The future of AI in drug development
AI is set to become even more important in the drug creation process. More and more, pharmaceutical firms are using AI in their research and development (R&D) to make it faster and better. This merger can happen in a number of ways, such as through internal growth, relationships, and deals. As AI systems get better, these models will probably be able to make even better predictions and work even better. Biotech companies that focus on technology and pharmaceutical companies that have been around for a while are also likely to work together more. This agreement brings together experts in both technology and drug research, which will speed up the process of making new treatments. AI is also important for the rise of personalized medicine because it makes it possible to create treatments that are specific to each person’s genetic makeup. This means that treatments will work better and patients will have better results. Using AI in the pharmaceutical business could speed up the process of making new drugs, lower prices, and make treatments better. AI technology is expected to lead to new and better treatments in the future, which will help people all over the world.
