International collaboration for the study of drug action based on real-world clinical data

2023 Fiscal Year Final Research Report

• Project/Area Number: 18KK0216
• Research Category: Fund for the Promotion of Joint International Research (Fostering Joint International Research (B))
• Allocation Type: Multi-year Fund
• Review Section: Medium-sized Section 47:Pharmaceutical sciences and related fields
• Research Institution: Kyoto University
• Principal Investigator: Kaneko Shuji 京都大学, 医学研究科, 研究員 (60177516)
• Co-Investigator(Kenkyū-buntansha): 白川 久志 京都大学, 薬学研究科, 准教授 (50402798) ; 宗 可奈子 京都大学, 薬学研究科, 助教 (50816684) ; 永安 一樹 京都大学, 薬学研究科, 助教 (00717902)
• Project Period (FY): 2018-10-09 – 2024-03-31
• Keywords: リアルワールドデータ / 有害事象 / 化学構造 / 受容体親和性 / 機械学習 / ルールマイニング / 予測

Outline of Final Research Achievements

Regarding detection of drug interactions and discovery of adverse events using real world data (RWD), we developed a gold standard that covers adverse events consisting of 92 types of positive and negative control pairs. As a result of evaluating whether early detection of adverse events is possible by applying association rule mining to medical receipts, we succeeded in detecting adverse event signals earlier than conventional methods even when using monthly receipt data. For the prediction of the receptor affinity of compounds, we normalized the chemical structures and affinity data for 1.71 million compounds, applied machine learning to the relationship between chemical structural formulas and receptor affinity, and succeeded in predicting affinity for new chemical structures.

Free Research Field

薬理学、医療情報学

Academic Significance and Societal Importance of the Research Achievements

実臨床データからリアルタイムに有害事象や薬物相互作用を発見するための基礎的なプロトコルおよび評価基準を作成することに成功し、今後の臨床データの利活用が期待される。化合物の化学構造に基づく受容体親和性の予測についても精度の高い方策を見いだすことができ、創薬シーンへの活用が期待できる。