AMAZINGINDEX.COM 每天一点 Amazing
47.9
VOL. 2026.10
2026.10.23
← 返回 2026.10.23 日报
日报快照 · Daily Snapshot
NO. 013

AI 冠脉炎症筛查发现43%漏诊率

#ARTICLE HackerNews 2026.10.23
推荐指数 52.0 NO. 013 · 2026.10.23
发布2026/10/22Score51Comments29

Caristo 的 CaRi-Heart AI 结合 CCTA 影像,在无症状低风险人群中检出42.9%存在冠状动脉炎症,传统血检和心血管风险评估完全无法识别。对预防医学和高端体检赛道是实质性突破,可能改写心血管早筛标准。

这项研究真正的信号是 AI 影像正在从辅助诊断转向定义新的疾病标准。传统 Framingham 风险评分和 LDL-C 指标在这里失效了,意味着心血管风险评估范式可能出现裂缝。

对做医疗 AI 的团队来说,Caristo 的路径值得研究:它不跟影像科拼读片速度,而是创造一个新的生物标志物(炎症指数)来打开付费场景。高端体检和保险精算是最先受益的下游。

风险在于这是 Biograph 的会员制研究,样本选择和随访数据尚未公开,42.9%这个数字在更广泛人群中会被稀释多少,决定了这是真突破还是精准营销。

负面 28 条评论

核心争论:AI炎症指标是真实突破还是BMI的复杂包装

esalman

Access to the study requires $17,500 per year, just fyi.

pixl97

That or OpenAI needs to turn back on their highly persistent model and we ask it about coronary inflammation.

bonsai_spool

I don't see how one can spin this into a positive finding - their preprint says that the only correlate they find with their new measurement is BMI. By my reading, not even male sex is associated with more inflammation. It sounds like they've identified a fancy BMI estimator, and not a very good one

替代方案: hs-CRP血液检测CAC冠脉钙化评分PREVENT评分ASCVD评分ziltivekimab抗炎药
查看原文 →