산업동향
How a Top-10 Pharma Creates Clinical Trial Benchmarks with NLP
- 등록일2023-09-15
- 조회수2006
- 분류산업동향 > 종합 > 종합
-
자료발간일
2023-09-15
-
출처
IQVIA
- 원문링크
How a Top-10 Pharma Creates Clinical Trial Benchmarks with NLP
◈ QUICK FACTS
◈ Situation: Pharma is constantly looking for ways to optimize the design of clinical trials. A crucial area is benchmarking drug efficacy, which is critical to identifying how well a drug must perform in a trial to become a worthwhile treatment. The required information is locked in multiple unstructured text sources. NLP can unlock it by combining established and large language model (LLM) techniques to extract and structure this information for analysis and better-informed decision making.
◈ Solution: NLP was used to extract oncology efficacy endpoints from the literature. Applying a combination of NLP technologies (established rules-based methods, and LLMs), several models were developed, trained, and tested to extract the endpoints from unstructured text. The best performing model was then used in production.
◈ Success: Extracting drug efficacy endpoints from the literature is a laborious manual task that is prone to human . The best performing NLP model to extract endpoints with high precision and recall produced a 96% F-Score. The AI-driven literature retrieval successfully ensured that a cancer clinical trial plan had the appropriate benchmarks.
☞ 자세한 내용은 원문링크를 이용하시기 바랍니다.