GPT-4 在医疗摘要方面准确性有所提高，但仍会出现幻觉

作者 PulseAugur 编辑部 · [1 个来源] · 2026-09-10 10:06

一项研究评估了大语言模型在生成急诊科就诊摘要方面的有效性。与 GPT-3.5 Turbo 相比，GPT-4 表现出更高的准确性，但两个模型在事实一致性方面都存在问题，GPT-4 生成的摘要中有 42% 出现了幻觉，47% 的病例遗漏了相关的临床信息。 AI

影响大语言模型在医疗文档记录方面显示出潜力，但临床使用需要显著提高准确性和完整性。

排序理由该集群包含一篇评估大语言模型在特定任务上性能的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 12 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Tool; 该集群包含一篇评估大语言模型在特定任务上性能的研究论文。[lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: Breaking (< 6h); Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

完整方法见我们的编辑标准。

Mastodon — mastodon.social TIER_1 English(EN) · [email protected] · 2026-09-10 10:06

"评估大型语言模型以起草急诊科就诊摘要" GPT-4 比 GPT-3.5-turbo 更准确，但 42% 的摘要存在问题

"Evaluating large language models for drafting emergency department encounter summaries" GPT-4 was more accurate than GPT-3.5-turbo, but 42% of summaries had hallucinations and 47% omitted relevant clinical information. # LLM # AI # Healthcare https:// doi.org/10.1371/journal.pdi…