Pinned post
1/9 🧵🚩 In clinical AI, overall accuracy does NOT equal safety.
In our new study in @JACCJournals presented simultaneously at #AHA26, @RahulAggarwalMD and our ⭐ team show how strong overall LLM performance can conceal potentially harmful treatment recommendations.
In our new study in @JACCJournals presented simultaneously at #AHA26, @RahulAggarwalMD and our ⭐ team show how strong overall LLM performance can conceal potentially harmful treatment recommendations.