Bringing Medical Advances from the Lab to the Clinic
Keywords: Artificial Intelligence; Large Language Models; Exploration
The high-quality large language models developed domestically, such as DeepSeek, undoubtedly have significant potential impacts on the development of various fields.
Foremost among these is language editing.
Our team has been applying AI since the launch of ChatGPT, so we are naturally curious about DeepSeek. Thus, we conducted a series of explorations to investigate how our colleagues break the “language barrier” in academic writing.
Please write a review article titled “Emerging Biological Therapies for Myasthenia Gravis” for publication in The Lancet, approximately 8000 words, with references.
The reason for choosing this topic is that the author of this article has been engaged in research in this area and is qualified to evaluate it.
A. ChatGPT o3-mimi-high, developed by OpenAI in the USA; it is a paid version.
B. Grok 2, developed by Elon Musk’s xAI in the USA; it is a paid version.
C. DeepSeek DeepThink (R1), developed by China’s High-Flyer; it is a free version.
All three tested AIs are open for online retrieval. Unfortunately, DeepSeek still shows “(due to technical reasons, online search is temporarily unavailable).”
Therefore, we will report based on the actual results received:
Added a thought process, similar to DeepSeek, no wonder it is considered the o3 forced out by DP!
The English review written by ChatGPT o3-mini-high is very academic, with complete and specific content, meeting the requirements of “comprehensive, new, smooth” or “faithful, accessible, elegant” for review writing; it is written in an academic style;
Updated to 2025, providing seven references with clear citation formats, which we verified one by one, all completely accurate.
At the bottom of the completed review, Grok 2 noted that this review was written according to the format requirements of The Lancet, but the content is thin, far from the required 8000 words. However, it can serve as a framework.
The latest references are up to 2024.
Provided six references with clear citation formats, which we verified one by one, all accurate.
C. DeepSeek (DeepThink(R1))
DeepSeek first states: due to technical reasons, online search is temporarily unavailable.
Then, DeepSeek provided the framework and core content for the academic paper; it did not provide a complete paper like ChatGPT.
However, as a framework for writing a paper, it is still very complete and accurate.
Among the four references provided, we found the first two accurate; we could not find the corresponding literature for the last two.
We will continue to explore and find faster ways for DeepSeek to help colleagues eliminate the “non-native English” barrier in academic paper writing.
Follow us as we explore ways to use artificial intelligence to improve work efficiency!
Click below to follow AI in Medicine
Author: Dr. Mark。Assistant: ChatGPT; Grok; DeepSeek
Healsan Consulting, USA, specializing inHealsan Medical Big Data analysis (Healsan™), and based on big dataHanson Clinical Research training (HansonCR™) and Medical Editing Services (MedEditing™). Mainly provides analysis and reports for physician-scientists, biopharmaceutical companies, and hospital research departments, becoming many institutions’ “Clinical Research Plugin”.
Website: https://healsan.com/
For more exciting content, click the bottom left corner “Read the Original”.
Click above; From Bench to Bedside, Healsan Paves the Path.
▼ Clinical Research Skills and Literature Updates.
（Click the image above to enter the album you are interested in. Or click “Resources“, to browse all resources of this public account.）
