Search papers, labs, and topics across Lattice.
School of Computing and Information Systems, University of Melbourne, Melbourne, Australia
4
0
4
Token-level explanations reveal how FEMRs leverage patient history, bridging the gap between black-box models and clinical trust.
Self-harm prediction models falter across hospitals due to stark lexical differences in triage notes, revealing a critical gap in model generalizability.
Suicide memes reveal unique modeling challenges, with higher severity levels often underpredicted, highlighting critical gaps in content moderation strategies.
Achieving over 95% accuracy in identifying self-harm methods, this approach revolutionizes surveillance by moving beyond binary classifications.
