weipenghan.bsky.social
PhD candidate in Health Policy & AI | Hospital digitalization, patient experience
👤 Joined on 11 August 2026
Followers
Following
Total Posts
@weipenghan.bsky.socialLeaving clinical data analysis to AI agents risks injecting silent errors into medical research. In a study using data from 7,802 patients, an LLM agent generated near-identical Kaplan-Meier estimates across 17 runs, but still produced critical execution failures.
Show PostSpot on. Local inference isn't just for privacy—it’s for reliability. Open source allows for the "last mile" of fine-tuning on local demographics without the HIPAA headache. Great build.
Show PostThis tiny magnetic e-reader snaps to the back of a smartphone. “By keeping it on the back of my phone at all times, I finished a 250-page book in two weeks (which usually takes me twice as long),” Brenda Stolyar writes. But the downsides outweigh the conveniences, she found.
Show Post