Hasty Briefsbeta

Bilingual

High-sensitivity pan-cancer AI assessment of lymph node metastasis via uncertainty quantification - PubMed

4 hours ago
  • #lymph node metastasis
  • #uncertainty quantification
  • #AI diagnostics
  • UPATHLN is a high-sensitivity AI platform for assessing lymph node metastasis across various cancers.
  • It uses a pathology foundation model-based encoder and a decoupled uncertainty estimation mechanism to improve accuracy.
  • Validated on 26,229 lymph nodes from 14 primary origins, it achieved an AUC of 0.986 in internal validation.
  • The uncertainty module flags potential false-negative predictions, ensuring 100% conditional sensitivity.
  • It reduces the review burden on negative lymph nodes by 73.2%.
  • UPATHLN sets a new benchmark for safety-critical AI diagnostics at the pan-cancer scale.
  • Authors have filed a patent related to the uncertainty-aware module and active learning framework.