[비즈한국] The use of artificial intelligence (AI) in drug development has now become essential. AI is being actively utilized in identifying drug candidate substances, designing clinical trials, and other processes. Insilico Medicine, a Hong Kong-based AI biotech firm, has stated that by applying AI technology across the entire process, it took only 46 days to identify a candidate substance and 30 months for that candidate to enter clinical trials.
Two End-to-End AI Candidate Substances Enter Phase 2 Clinical Trials
According to market research firm MarketsandMarkets, the global AI drug development market was valued at $902.7 million (1.3 trillion KRW) as of 2023, showing an average annual growth rate of 40.2%. It is projected to reach a scale of $4.8936 billion (7 trillion KRW) by 2028. The AI drug development market is largest in machine learning, followed by natural language processing, situational awareness processing and computing, and other technologies. By therapeutic area, the market size is largest in oncology, followed by infectious diseases, neurology, cardiovascular diseases, immunology, metabolic diseases, and others. Specifically, as of 2028, oncology is estimated to reach $1.947 billion (2.7 trillion KRW), and infectious diseases are estimated to reach $1.094 billion (1.5 trillion KRW).
Companies utilizing AI for drug development began emerging in the late 2000s, followed by partnerships between AI firms and pharmaceutical companies or the establishment of dedicated in-house AI organizations. According to the Korea Health Industry Development Institute, there were 18 joint research and partnership cases related to drug development last year alone. Recently, tangible results have emerged, such as Insilico Medicine and Recursion (USA) advancing candidate substances developed through end-to-end AI into Phase 2 clinical trials. In addition, four other companies, including South Korea’s Oncocross, which targets muscle diseases such as sarcopenia, have drug candidates in Phase 1 clinical trials.
Clinical Success Rates Using AI Exceed Industry Averages
The reason for using AI in drug development is simple: to reduce time and costs. The drug development process consists of: target identification, lead compound discovery and optimization, preclinical trials, and clinical trials. AI is being used more to complement rather than replace specific stages. So far, it is primarily used in identifying drug candidates and designing clinical trials. Insilico Medicine stated that by using AI, the candidate discovery process was shortened to 46 days, the time until the candidate entered clinical trials was 30 months, costs were cut by 90%, and time was reduced by two-thirds. Typically, it takes 10–12 years for a candidate substance to receive new drug approval, costing about 3 trillion KRW.
The high success rate of using AI is also reflected in statistics. According to a paper written by Madura KP Jayatunga’s research team, 75 substances discovered via AI have entered clinical trials in the last 10 years, and in the case of oncology, about 50% of substances in Phase 1 and 2 were discovered through AI. A preliminary analysis of clinical trial success rates shows that as of December 2023, 24 substances had completed Phase 1 trials, with 21 of them succeeding. For Phase 2, 10 substances completed trials, of which 4 succeeded. In their paper, the research team stated, “Phase 1 showed an 80–90% success rate and Phase 2 showed a 40% success rate, which are similar to or higher than the industry averages of 40–65% and 30–40%, respectively.”
Domestic companies are also actively utilizing AI. JW Pharmaceutical001060 launched its AI-based drug research and development integration platform, JWave (JW AI-powered Versatile drug Exploration), last year. JWave is a platform that integrates 'JWELRY' and 'CLOVER,' the big-data-based AI drug systems the company previously operated. It can utilize genomic information from over 500 cell lines, organoids, and various disease animal models for training. Daewoong Pharmaceutical069620 developed 'AIVS (AI based Virtual Screening),' which can be applied to the first stage of drug candidate discovery, and 'DAISY (Daewoong AI System),' an AI drug development portal, last year. DAISY allows for ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) research, a research stage that determines a compound's drug-like properties.