建構生物醫學文獻與知識庫探勘技術以尋找舊藥之新用途

Abstract

新藥物的推出能為藥品公司帶來相當的收入。然而，典型的新藥開發需要耗費大量資金與時間，且通過實驗、順利上市的成功率相當低，因此大部分的投資無法回收。近年來，許多藥品公司逐漸引入「舊藥新用」作為藥品開發的替代研發方法。「舊藥新用」是從既有的藥物，在其原本設計標的之外，尋找新適應症的藥物開發方式；由於既有藥物已經有許多前置研究基礎，可以省去許多臨床前評估與測試，藥品公司因而可以減少開發的時間與資金成本。 本研究基於Swanson提出的文獻探勘方法，分析超過15,000,000篇生物醫學文獻、以及藥物與疾病之知識庫，以自動化尋找尚未被發現且可能有直接關聯的既有藥物與疾病關係。我們建立三個實驗情境以評估本研究所提出之方法效能，其結果顯示，本研究提出之方法與所建構之綜合生物醫學概念網路能有效較既有方法有效提供潛在的藥物與疾病關係給研究者，以幫助研究者尋找可能的舊藥之新用途。

Drug development is time-consuming and costly. However, most of drug development projects fail before they ever enter into clinical trials. To reduce the high risk of failure for drug development, pharmaceutical companies are exploring the drug repositioning approach for drug development. Previous studies have shown the feasibility of using computational methods to help extract plausible drug repositioning candidates, but they all encountered some limitations. We thus propose a novel drug-repositioning discovery method that takes into account multiple information sources, including more than 15,000,000 biomedical research articles and existing ontologies that cover detailed information about drugs, proteins and diseases, and follow the ABC model derived from Swanson’s literature-based discovery works. We design three experiments to evaluate our proposed drug repositioning discovery method. The results show that our proposed method and our proposed integrated information source can better help researchers sift plausible drug-disease relationships in comparison with existing techniques.