Unverified50% confidenceBenchmarkExact time
The offline entailment proxy model built by the researchers can distinguish reports written for different patients with an AUC of 0.987.
1
Sources
50%
Confidence
Long-term
Relevance
9/15/2026
First Seen
Sources
Related Entities
Related Claims
Unverified在MMLU医学子集上得分优异的模型，在处理真实ICU病历时可能表现平平，因存在分布偏移（Distribution Shift）62% similarUnverified在实测中，AI Agent对心脏病预测数据集识别出303条样本、13个特征以及一个二分类标签62% similarUnverifiedA meta-analysis published in the Journal of Medical Internet Research in 2022 confirmed that automated follow-up resulted in patient response rates 40–60% higher than those of groups with no follow-up.61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/921813API
curl https://kongchang.com/api/v1/knowledge/claims/921813MCP
get_claim(id=921813)