Generative AI Driven Clinical Drug Development
Shinichi Tsuchiwata, Yasuhisa TANABE, Yumiko HOSOYA, T Yonekura, Miki Someya, Kyoko ITAMI, Ayami KITAMURA, Misaki ISHII, Yamei HAN, Hisanaga OHASHI, Riko Tabuchi, Takumi Tanaka, Taku Uryu, Mikiko Noyes, Noriko Matsumoto, Yuji Isobe, Tomoka SAOTOME, Kazuki MIMURA, K. Tsujimura, Ryoichi NAKAMURA · Rinsho yakuri/Japanese Journal of Clinical Pharmacology and Therapeutics · 2025
Generative Artificial Intelligence (AI) is anticipated to enhance productivity across all industries. This paper investigates the application of generative AI in the field of new drug development, focusing on five areas of use: data processing, text analysis, documentation for professionals (e.g. investigators and regulators), generating meeting minutes, and documentation for clinical trial participants. In data processing, generative AI can be used to assist in the creation of analysis programs (codes) and to interpret codes created by others. In text analysis, the application of generative AI to unstructured data collected from free-text responses, such as questionnaire answers, enables classification and information extraction. The scope of generative AI application in documentation is diverse, facilitating efficiency in translation, summarization, searchability, structure modification, information updating, and review of professional texts. Furthermore, employing generative AI that can refer to technical terms and company's abbreviations can result in precise meeting minutes in a shorter time. Generative AI can also support to create documents that are easy for patients and patient's family to understand and to generate anticipated questions from a patient's perspective about clinical trials, making it a useful communication tool with trial participants. The application of generative AI in drug development is believed to enable efficiency in various situations. However, it is important to understand the advantages and limitations of generative AI and to find appropriate application and operational methods.