%0 Journal Article
%T 基于“DeepSeek + 医疗”在线评论的情感分析
Sentiment Analysis of “DeepSeek + Healthcare” Online Reviews
%A 曲兆悦
%J Statistics and Applications
%P 103-115
%@ 2325-226X
%D 2025
%I Hans Publishing
%R 10.12677/sa.2025.146151
%X 近年来，人工智能在医疗方面的应用越来越广泛，2025年1月掀起了国产大模型DeepSeek的热潮，全国多家医院陆续接入了DeepSeek技术，如西安国际医学中心医院，该技术的接入可以提升诊断效率及就医体验。随着DeepSeek大模型在医疗领域广泛部署，其应用效果引发关注，但针对用户情感态度与使用意愿的系统研究仍然缺乏。为此，本研究采用Python爬虫从抖音平台采集4000余条评论，经数据清洗和分词处理，使用TF-IDF提取文本特征，构建Transformer情感分析模型，并运用LDA模型提取评论主题。结果表明，评论情感以正面为主，多数用户认可“DeepSeek + 医疗”的诊断准确性与便利性，但少量负面评论涉及隐私、安全及AI可靠性顾虑。LDA分析识别出智能医疗、就医服务、医疗变革、诊疗方案四个主要主题。据此提出提高诊断精准度、优化用户界面和加强隐私保护等建议，以提升用户信任度和使用意愿。
In recent years, the integration of artificial intelligence (AI) into the healthcare sector has advanced rapidly. In January 2025, the large domestic language model DeepSeek garnered widespread attention, with numerous hospitals across China—including Xi’an International Medical Center Hospital—successively adopting this technology. The implementation of DeepSeek has demonstrated the potential to enhance diagnostic efficiency and improve the overall patient experience. Despite its growing deployment, there remains a paucity of systematic research on user sentiment and adoption intentions related to this technology. To address this research gap, this study employed Python-based web scraping techniques to collect over 4,000 user comments from the TikTok platform. Following data cleaning and Chinese word segmentation, textual features were extracted using the Term Frequency–Inverse Document Frequency (TF-IDF) method. A transformer-based sentiment analysis model was then constructed to assess public sentiment, and a Latent Dirichlet Allocation (LDA) model was utilized to identify underlying thematic structures within the comments. Findings reveal that public sentiment is predominantly positive, with the majority of users expressing approval of the diagnostic accuracy and convenience offered by the “DeepSeek + healthcare” model. Nonetheless, a minority of negative comments reflect concerns pertaining to data privacy, system security, and the overall reliability of AI technologies. The LDA topic modeling identified four principal themes: intelligent healthcare, medical service delivery, healthcare transformation, and diagnostic strategies. Based on these insights, the study proposes targeted recommendations to enhance diagnostic precision, improve user interface design, and strengthen data privacy safeguards, thereby fostering greater user trust and increasing the willingness to adopt AI-driven healthcare solutions.
%K DeepSeek + 医疗，
%K 人工智能，
%K 大语言模型，
%K 情感分析，
%K LDA主题模型，
%K 用户使用意愿
DeepSeek + Healthcare
%K Artificial Intelligence
%K Large Language Model
%K Sentiment Analysis
%K LDA Topic Modeling
%K User Adoption Intentions
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=118315