What AI offers drug repurposing  


DDW Editor Reece Armstrong speaks to Jouhyun Jeon, Principal Investigator at Klick Applied Sciences, about the company’s artificial intelligence algorithm, LOVENet, and how it offers a fresh perspective for repurposed drugs.  

RA: First off, could you tell me about the LOVENet algorithm? 

JJ: LOVENet (a Large Optimized Vector Embeddings Network) is the artificial intelligence framework we designed to address the challenges of identifying new therapeutic indications for existing drugs. It seamlessly integrates advanced machine-learning methods with extensive biological and clinical datasets to offer a fresh perspective on new potential therapeutic applications. 

RA: How can this algorithm – which can mathematically represent the relationships between drugs and diseases – offer insights into therapeutic applications? 

JJ: LOVENet could greatly improve the drug repurposing process and transform the pharmaceutical industry because it integrates two cutting-edge AI technologies to mathematically represent the relationships between drugs and diseases and offer a fresh perspective on new potential therapeutic applications. 

RA: Why is this algorithm important for drug repurposing strategies?  

JJ: LOVENet adopts two technologies – knowledge graph and large language model (LLM). The knowledge graph helps us to represent, analyse, and translate the relationships between drugs and diseases. The LLM utilises its capacity to infer drug-disease associations from extensive textual data. In particular, we converted the mathematical relationship between drugs and diseases into text format so that LLM can be trained. This combination allows for a more comprehensive analysis of drug-disease relationships by leveraging both structured (knowledge graph) and unstructured data (text data).  

RA: How much potential is there to repurpose existing drugs on the market? 

JJ: We believe there is significant market potential here given many existing drugs have secondary actions or effects that can be therapeutically relevant for conditions other than their original indication. According to several reports1, about 30-40% of current FDA-approved drugs or biologics are considered repurposed or repositioned products.   

RA: Following that, do you think the industry is missing out on untapped therapies that could be used in other treatment indications? 

JJ: About 20% of prescriptions are for an off-label use2. In certain subpopulations of patients, this rate may be even higher (some reports say up to 97%). When we consider this common off-label usage of drugs, there are likely numerous untapped therapies among existing medications that could be effective for other diseases. That is why many pharmaceutical companies are investigating computation-driven methodology to find novel indications for existing drugs and other drug development to reduce investigating time and labour efforts.  

RA: Can AI be used to shorten the development timeline for new medicines? 

JJ: The usual or typical path for developing new medicines can take more than a decade. Our work with LOVENet demonstrates how AI can speed up the repurposing process, as well as drug discovery. AI can ingest and process large volumes of complex biomedical data and help researchers or scientists to discover meaningful patterns (or clues) for drug discovery faster. And, as AI technologies develop, they are expected to become more adept at predicting clinically viable treatments. 

RA: What are the current repurposing methods and how can AI speed this process up? 

JJ: Current repurposing methods include two approaches. 

The first is an experiment-driven approach through the use of high-throughput drug screening. This approach uses automated equipment to rapidly test thousands to millions of drug molecules for biological activity at the model organism, cellular, pathway, or molecular level. While this approach significantly reduces research investigating time, it is costly and has a high failure rate.  

The second approach is computational drug discovery, which adopts AI and machine-learning technologies. While this approach is still considered experimental and in exploration, AI algorithms proved that it can analyse vast datasets to identify potential new uses for existing drugs. The beauty of AI is its ability to handle large-scale data analysis and find hidden patterns that can represent repurposing drugs, resulting in improvement of success rate.  

RA: In terms of AI’s current use within drug discovery, we’ve seen it used within the discovery space quite well, but therapies have still failed to make it to market due to poor translatability. Is this something you’re concerned about or is it still early days for the technology within pharma? 

JJ: While AI has accelerated the discovery phase, many AI-driven therapies fail in clinical trials due to issues like efficacy, safety, or unexpected side effects.   

However, we believe that as AI methods become more sophisticated and integrate more  

comprehensive biological data – including clinical/pathological data, the translatability will improve. Researchers are still investigating and collecting biological and clinical data to get the comprehensive and systematic view of biological knowledge, international level collaboration, and unified protocol to collect and process data that’s required. It is still relatively early in the integration of AI into pharma. As the technology evolves, so will its efficacy in predicting clinically successful therapies.  


Jouhyun Jeon holds a PhD in computational biology. With a background in academic research and postdoctoral training, she has developed machine-learning pipelines to identify anti-cancer therapeutics and built predictive models for drug and biomarker discovery. As a lead of data-driven product development at Klick Applied Sciences, Jeon tackles medical challenges by innovative methodologies. 


1: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9336118/

2: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9998554/

Related Articles

Join FREE today and become a member
of Drug Discovery World

Membership includes:

  • Full access to the website including free and gated premium content in news, articles, business, regulatory, cancer research, intelligence and more.
  • Unlimited App access: current and archived digital issues of DDW magazine with search functionality, special in App only content and links to the latest industry news and information.
  • Weekly e-newsletter, a round-up of the most interesting and pertinent industry news and developments.
  • Whitepapers, eBooks and information from trusted third parties.
Join For Free