230802 리뷰논문쓰기 브이로그 1일차 - 리뷰논문 작성시작, ChatGPT로 구조 잡기, 인용논문 찾기

English summary

The video documents the workflow for drafting a review paper on applications and possibilities of large language models in clinical trials. It describes using Obsidian to search, summarize, synthesize, and structure evidence, together with ChatGPT and citation-network tools to refine the outline and organize references.

Key points

• Obsidian is used for the paper outline, topic notes, daily work logs, and links among references and concepts. The workflow also separates verbatim quotations from paraphrased and interpreted text to reduce unintended plagiarism.
• ChatGPT GPT-4 is used to assess the review structure, suggest additions such as summary or comparison tables in the conclusion, and correct formatting problems in text copied from PDFs.
• Perplexity AI, Litmaps, Connected Papers, ResearchRabbit, Zotero, Google Scholar, PubMed, arXiv, and SciSpace support literature searching, citation-network exploration, bibliographic management, PDF retrieval, rapid summarization, and targeted questions. Very recent papers may not yet be fully indexed by these tools.

Audience

The content is intended for researchers, medical and clinical informatics specialists, and advanced students interested in literature-review workflows and medical AI evidence discovery.

Scope and limitations

This source is a video record of one researcher’s review-writing workflow and tool use. The reported examples involving trial matching, medical coding, protocol design, and medical-text simplification are presented as findings identified in the reviewed literature; the video itself does not independently validate model performance. The source explicitly notes incomplete indexing of recent papers, failures of AI summarization tools when context is absent, and the risks of hallucination, limited domain knowledge, and privacy concerns. It states that expert human oversight remains necessary for medical deployment.

Research context

The topic has a thematic connection to medical AI and data science through its focus on using large language models for clinical-trial literature discovery, information extraction, medical classification, and clinical-document processing.