HONG KONG – The Korean 2020 KoNECT-MOHW-MFDS International Conference, which is taking place online this year due to the COVID-19 pandemic, faced a challenging start.
Technical difficulties hampered some of the early proceeding. However, technology took focus again later in the day in a more positive light, with an exploration of artificial intelligence (AI).
Seonee Nam, leader of SK C&C’s AI-based drug discovery platform services team, kicked off her talk by identifying digital transformation as a “new wave” in drug research and development. In terms of drug R&D productivity, she said that costs are increasing year-on-year, but the number of new drug approvals is not seeing the same increase.
R&D costs reached record levels of $2.168 billion in 2018, but R&D returns saw a record low of 1.9% in the same year, she said, citing Deloitte’s "Unlocking R&D productivity: Measuring the return from pharmaceutical innovation 2018" report.
In order to speed up and improve the efficacy of the drug development process, Nam identified drug discovery as the “obvious” phase most likely to benefit from digital transformation, as she cited limits to the level of digitization possible during the clinical trials phase. Although not all of the drug discovery process can be digitalized, shifting from a wet lab-centric focus to a dry and wet lab-centric one would offer the speed and efficiency of the discovery process via the quick identification of candidate targets from the large amount of data available, effective candidate compound design, fast verification of candidate targets as well as compounds, and quick access to the latest data.
Nam identified three specific parts of drug discovery - assay development to identify molecules, lead identification and lead optimization - as the areas that “many companies have prioritized digitizing to save time, become more cost efficient and increase their success rate compared to the wet lab-centric model,” she said.
Target discovery was also Korean pharmaceutical companies’ biggest weak point, according to Dong-A Ilbo, a Korean newspaper that surveyed the top 30 pharmaceutical companies in the country in May 2016. Challenges include building a big, diverse literature collection and database inclusive of a full set of scientific articles published by corresponding experts; finding and prioritizing novel disease-related targets.
SK C&C’s iCLUE-ASK system, being co-developed with Standigm, is one tool being developed to address the challenges. The approach involves constructing a database based on natural language processing technology, generating a heterogenous knowledge graph, and applying algorithms for target identification, according to the company.
Another tool that SK C&C is developing is the iCLUE-TDMD target and phenotype-based disease model, developed with inspiration from David Swinney and Jason Anthony’s article “How were new medicines discovered?” published by Nature Reviews Drug Discovery in June 2011. Data in the article said that 51% of follow-up drugs were discovered via target-based screening, compared to 57% of first-in-class drugs being discovered by phenotypic screening.
Wrapping up, Nam said that AI could also be of use in clinical trials, where it could be used during trial design, by using already-available data to design primary and secondary endpoints. She also pointed out patient recruitment as another area that could be enhanced by the use of the technology.